What judges actually weigh when they decide custody

← Back to articles

Every state's custody statute invokes some version of the "best interests of the child" standard. Few legal phrases are repeated more often, and few are understood more loosely by the parents whose families it governs. This article is a careful look at what the standard actually contains, how it evolved, the factors judges typically weigh, the documented criticisms of the framework, and the protections that have been built in for survivors of domestic violence.

Where the modern standard comes from

Before the twentieth century, English common law treated children as essentially the father's. Through the nineteenth and early twentieth centuries, equity courts shifted toward an explicit focus on the child's welfare, articulated most influentially by Judge Cardozo in Finlay v. Finlay, 240 N.Y. 429 (1925). For much of the twentieth century, the "tender years" doctrine gave a strong presumption to mothers, particularly for young children, until equal- protection challenges following Orr v. Orr, 440 U.S. 268 (1979), eroded that presumption.

The modern multi-factor framework was nationalized by the Uniform Marriage and Divorce Act, promulgated in 1970 and amended in 1973. UMDA section 402 reads: "The court shall determine custody in accordance with the best interest of the child. The court shall consider all relevant factors including: (1) the wishes of the child's parent or parents as to his custody; (2) the wishes of the child as to his custodian; (3) the interaction and interrelationship of the child with his parent or parents, his siblings, and any other person who may significantly affect the child's best interest; (4) the child's adjustment to his home, school, and community; and (5) the mental and physical health of all individuals involved."

UMDA section 402 was a deliberate repudiation of the tender-years presumption. It directed that no single factor be dispositive, that gender be irrelevant, and that the court weigh a holistic constellation of circumstances. All 50 states, the District of Columbia, and U.S. territories now apply some version of a best- interests framework. Roughly 31 states and D.C. enumerate specific statutory factors (Children's Bureau, Determining the Best Interests of the Child, 2023).

The typical factors

The Children's Bureau's 2023 survey of state statutes identifies a recurring core of factors. Different states organize and weight them differently, but the following appear across most jurisdictions.

The wishes and capacity of each parent. Proposed parenting plans, demonstrated commitment, the capacity and disposition to act on the child's needs as opposed to the parent's.
The wishes of the child. Age- and maturity- dependent. No state allows a minor to simply choose, except in a narrow Georgia provision discussed below.
Existing relationships. The interaction with each parent, with siblings, and with other people who significantly affect the child. Twenty-two states and D.C. codify emotional ties as a discrete factor. Sibling unity is a near-universal judicial preference.
Adjustment to home, school, and community. Stability and continuity. Academic performance and school consistency. The geographic viability of proposed plans.
Mental and physical health of all parties. Twelve states and D.C. enumerate the parents' health; 15 enumerate the child's. Diagnosed mental illness is generally protected under the ADA unless it demonstrably impairs parenting. Substance abuse is considered in most jurisdictions.
History of domestic violence. Universally considered, codified as a discrete factor in 12 states and addressed through statute or case law in all 50.
Willingness to facilitate the relationship with the other parent. The "friendly parent" doctrine, codified in many states and now narrowed by domestic violence carve-outs.

Four state statutes worth looking at

Four representative statutory frameworks illustrate how the same general standard is implemented very differently.

Michigan: the classic 12-factor model

Mich. Comp. Laws section 722.23, the Child Custody Act of 1970, is widely regarded as the most complete codification of the UMDA framework. "Best interests of the child" means the sum total of 12 factors: love and affection and emotional ties; capacity and disposition to give love, affection, guidance, and continued religious education; capacity to provide food, clothing, medical care; length of time in a stable, satisfactory environment and the desirability of maintaining continuity; permanence as a family unit; moral fitness; mental and physical health; the child's home, school and community record; the reasonable preference of the child if of sufficient age; willingness and ability of each party to facilitate a close and continuing parent-child relationship with the other parent (with an explicit carve-out that a parent may not be penalized for reasonable action taken to protect against domestic violence or sexual assault); domestic violence regardless of whether directed at or witnessed by the child; and any other factor the court considers relevant. Michigan courts must explicitly address each factor in written findings. Failure to do so is reversible error.

Florida: 20 factors and an equal-time presumption

Fla. Stat. section 61.13(3) enumerates 20 factors, the most comprehensive list of any U.S. state. As of 2023, Florida law also contains a rebuttable presumption that equal time-sharing is in the child's best interests. A party seeking deviation must prove by a preponderance that equal sharing is not. The statute requires courts to specifically acknowledge in writing any evidence of domestic violence, sexual violence, child abuse, child abandonment, or neglect. A domestic violence conviction creates a rebuttable presumption that shared parental responsibility is detrimental. Knowingly providing false domestic violence information to the court is itself a negative factor in the best-interests analysis.

California: streamlined, with safety as paramount

Cal. Fam. Code section 3011 takes a more streamlined approach, identifying as primary factors the health, safety and welfare of the child; a history of abuse against the child, the other parent, or any person in the household; the nature and amount of contact with both parents; and habitual or continual illegal substance use or alcohol abuse (subject to independent corroboration). Section 3011(b) explicitly prohibits considering the sex, gender identity, gender expression, or sexual orientation of any parent or guardian. The child's wishes are addressed separately in Family Code section 3042, which provides that a child 14 or older shall be permitted to address the court unless the court finds it not in the child's best interests.

New York: discretionary, case-law-driven

New York is distinctive because it does not enumerate a statutory checklist. N.Y. Dom. Rel. Law section 240 directs courts to enter custody orders "as, in the court's discretion, justice requires." The substantive framework comes from case law, most notably Eschbach v. Eschbach, 56 N.Y.2d 167 (1982), which identifies quality of home environment, ability to provide for emotional and intellectual development, financial capacity, relative fitness, the effect of an award on the child's relationship with the other parent, the child's preference (weighted by maturity), and stability. New York's framework is more openly discretionary than Michigan's or Florida's, which makes outcomes more flexible and, the criticism runs, more unpredictable.

Children's voices: how courts hear them

No state allows a minor to simply choose a custodian, with one near- exception. Georgia's O.C.G.A. section 19-9-3 provides that a child aged 14 or older may elect the parent with whom they wish to live, creating a strong (though rebuttable) entitlement that a court may override only by finding it would not serve the child's best interests. Most states use a "sufficient age and capacity" standard, leaving the judge to assess maturity and reasoning case by case. A few set named thresholds: Tennessee requires mandatory consideration at age 12, Indiana gives enhanced weight at 14, California's procedural rule permits a child 14 or older to address the court.

Procedurally, courts hear children's preferences through in-camera judicial interviews (judge alone with the child, which protects from courtroom adversarialism), guardian ad litem testimony, court- appointed psychologist or custody evaluator reports, and the child's own attorney in jurisdictions like New York that use the attorney- for-the-child model.

The friendly-parent doctrine and its critics

The friendly-parent doctrine asks the court to favor the parent more willing to support the child's relationship with the other parent. It is codified in many state statutes, including Florida's section 61.13(3)(a), Michigan's MCL 722.23(j), and Virginia's section 20-124.3(6) to (7). The theoretical rationale draws on the empirical finding that children with strong relationships with both parents tend to fare better post-separation.

The doctrine has attracted serious academic criticism. Margaret K. Dore's 2004 article in the Loyola Journal of Public Interest Law, "The Friendly Parent Concept: A Flawed Factor for Child Custody," identifies three structural defects. The litigation paradox: in a custody dispute, each parent is by definition seeking to reduce the other's access, so simultaneously penalizing parents for appearing "unfriendly" creates an irrational dynamic. The safety paradox: a battered parent who seeks a protective order, objects to unsupervised visitation, or refuses to facilitate contact will be rated unfriendly under the doctrine, systematically disadvantaging domestic violence survivors. Joan Zorza documented this pattern in 1992 in the Clearinghouse Review. And the unpredictability: friendliness is subjective and context-dependent, producing wide divergence across judges and jurisdictions.

Lisa A. Tucker's 2022 article in the Fordham Law Review, "Domestic Violence as a Factor in Child Custody Determinations: Considering Coercive Control," argues that courts trained exclusively on physical violence miss coercive control. Because coercive control leaves no visible bruises, abused parents may appear uncooperative when they are in fact acting protectively.

A number of states have responded with explicit carve-outs from the friendly-parent factor where domestic violence is proven. Michigan's MCL 722.23(j): "A court may not consider negatively for the purposes of this factor any reasonable action taken by a parent to protect a child or that parent from sexual assault or domestic violence." Alaska Stat. section 25.20.090(6)(E), Oregon Revised Statutes section 107.137(1)(f), and Vermont Statutes title 15 section 665(b)(5) have similar provisions. Approximately 20 states maintain rebuttable presumptions against awarding sole or joint custody to a perpetrator of domestic violence.

What the Children's Bureau survey shows nationally

The Children's Bureau's Determining the Best Interests of the Child, current through September 2023, provides the most authoritative federal survey of state law in this area. A few findings worth noting:

Most states have no statutory definition of "best interests." Only Montana and Puerto Rico provide one. This absence of definition is a structural source of the indeterminacy criticism discussed below.
Roughly 32 states plus territories codify family integrity or a preference for keeping the child in the family home as a guiding principle.
31 states and the Northern Mariana Islands codify health, safety, and protection of the child as guiding principles.
Six states and Puerto Rico explicitly prohibit specific considerations: California (sex, gender identity, gender expression, sexual orientation), Connecticut (socioeconomic status of birth parent or caregiver), Delaware (gender presumption), Iowa and Minnesota (race, color, national origin), and Puerto Rico (a broad anti-discrimination list).

The indeterminacy critique

The most influential scholarly critique of the best-interests standard is Robert Mnookin's "Child-Custody Adjudication: Judicial Functions in the Face of Indeterminacy," 39 Law and Contemporary Problems226 (1975), with a 2014 retrospective in the same journal. Mnookin argued that the standard is indeterminate in three ways.

First, empirical uncertainty. Courts are asked to predict which arrangement will best serve the child's long-term development. The behavioral sciences, in 1975 and largely still today, cannot reliably predict how a particular child will fare under alternative arrangements. Causal links between specific parenting configurations and specific developmental outcomes are probabilistic, contested, and highly individual.

Second, value pluralism. Even with accurate prediction, choosing among outcomes requires a value judgment about what kind of childhood is best. Is intellectual stimulation more important than emotional stability? Economic security or religious upbringing? Courts have no neutral, shared framework for resolving these conflicts. The standard gives the judge wide discretion to project their own values onto the assessment.

Third, systemic cost. An indeterminate standard encourages litigation rather than settlement (because neither party can predict outcomes), produces inconsistent results that undermine rule-of-law values, invites bias to operate covertly under the cover of individualized discretion, and inflicts disproportionate transaction costs on litigants who can least afford them.

The field has, in the half-century since, moved toward a kind of structured discretion. Detailed statutory factor lists like Michigan's MCL 722.23 are an attempt at a middle ground: preserving the individualized judgment Mnookin acknowledged was unavoidable, while channeling judicial reasoning into transparent, reviewable categories. The Michigan Law Review's analysis of this tension, Discretion, Rules, and Law: Child Custody and the UMDA's Best- Interest Standard, is a useful entry point for the ongoing debate. In his 2014 retrospective, Mnookin acknowledged that the core indeterminacy critique remains valid but that it has to be weighed against the costs of any alternative rule. Rigid rules solve the inconsistency problem but create their own injustices.

What this means for a parent

A few practical implications.

The best-interests standard does give judges substantial discretion. That discretion is, however, channelled in most states by a fairly well-defined set of factors. Knowing the specific factor list in your jurisdiction is one of the highest-leverage pieces of preparation a parent can do. In a state like Michigan, where each factor must be addressed in writing, your case essentially has 12 separate questions to answer. In a state like New York, the framework is more discretionary but the case law identifies recurring themes that good counsel can address.

The status quo bias is real and substantial in most jurisdictions. Courts heavily weight the stability and continuity of a child's current placement. This cuts in different directions for different parents, and it is worth understanding rather than ignoring.

The friendly-parent factor exists and is taken seriously. So do the domestic violence carve-outs from it. If you are a survivor whose protective behavior could be misread as uncooperativeness, counsel who understands the relevant statutory carve-out, and how to put it in front of the court, is essential.

Documentation matters more than rhetoric. Each statutory factor that gets evaluated benefits from contemporaneous, verifiable evidence, not from after-the-fact characterization. Records of parenting involvement, school records, medical records, financial records, communications, and a clean and tamper-evident communication log can all carry weight under multiple factors at once.

Finally, the standard's indeterminacy cuts both ways. It means that outcomes are less predictable than parents would like. It also means that no parent's situation is foreclosed by a single bad fact, and that the architecture parents build (parenting plans, communication habits, the record of their own conduct) genuinely matters to how a court will see them. The standard is imperfect. It is also the standard. Understanding what it actually contains is the first step toward being able to operate inside it.

Published April 2026.