Importance: Homicide is a leading cause of death among children in the US. Objective: To examine trends in child homicide rates and characteristics most commonly associated with these deaths. Design, Setting, and Participants: In this cross-sectional study, the study team used National Vital Statistics System WONDER mortality data for 38 362 homicide victims aged 0 to 17 years for 1999 to 2020 and National Violent Death Reporting System data for child homicide victims for 2003 to 2019 in 45 states, Washington, DC, and Puerto Rico. WONDER data are based on death certificates for US residents. National Violent Death Reporting System data include characteristics of violent deaths, linking information from death certificates, and law enforcement, coroner, and medical examiner reports. Exposures: Child homicide was defined using underlying cause-of-death codes U01 to U02, X85 to Y09, and Y87.1 from the International Classification of Diseases, Tenth Revision, Clinical Modification. Main Outcomes and Measures: Trends in homicide rates per 100 000 children were tested using joinpoint regression analysis; differences in rates from 2019 to 2020 were evaluated using z tests. Circumstances of child homicides were described. Results: This study included 38 362 homicide victims (69.4% male). The overall child homicide rate (per 100 000 children) has increased annually, on average 4.3% since 2013, with a precipitous rise from 2019 to 2020 (2019 rate, 2.2; 2020 rate, 2.8; overall increase of 27.7%). Homicide rates recently increased significantly for boys (2018 rate, 2.9; 2020 rate, 4.1; overall increase of 16.1%), 6- to 10-year-olds (2014 rate, 0.5; 2020 rate, 0.8; overall increase of 5.6%), 11- to 15-year-olds (2018 rate, 1.3; 2020 rate, 2.2; overall increase of 26.9%), 16- to 17-year-olds (2018 rate, 6.6; 2020 rate, 10.0; overall increase of 19.0%), Black children (2012 rate, 5.9; 2018 rate, 6.8; 2020 rate, 9.9; overall increase of 16.6% from 2018 to 2020), Hispanic children (2014 rate, 1.6; 2020 rate, 2.2; overall increase of 4.7%), children in the South (2013 rate, 2.1; 2020 rate, 3.5; overall increase of 6.4%), and in rural (2011 rate, 1.8; 2020 rate, 2.4; overall increase of 3.2%) and urban areas (2013 rate, 1.9; 2020 rate, 2.9; overall increase of 4.4%). Since 1999, homicide rates have decreased for girls (1999 rate, 1.9; 2020 rate, 1.5; overall decrease of 1.4%), infants (1999 rate, 8.7; 2020 rate, 6.6; overall decrease of 1.3%), 1- to 5-year-olds (1999 rate, 2.1; 2020 rate, 1.8; overall decrease of 1.0%), Asian or Pacific Islander children (1999 rate, 2.0; 2020 rate, 0.5; overall decrease of 4.4%), White children (1999 rate, 1.5; 2020 rate, 1.3; overall decrease of 0.7%), and children in the Northeast (1999 rate, 2.0; 2020 rate, 1.7; overall decrease of 1.4%). Homicides of children 10 years or younger were most commonly precipitated by abuse/neglect, perpetrated by parents/caregivers. Homicides of 11- to 17-year-olds were most commonly precipitated by crime and arguments and perpetrated by someone known to them, especially friends and acquaintances. Conclusions and Relevance: The decline in homicide rates for some geographic and child demographic groups is encouraging; however, rates recently increased across several subpopulations, with some racial and ethnic disparities persisting for more than 20 years. More targeted strategies are needed to (1) protect 6- to 10-year-olds, 11- to 17-year-olds, and children in certain geographic areas and (2) urgently address firearm violence, racism, and inequities at the root of youth violence.