From an early age, most children choose to play with toys typed to their own gender. In order to identify variables that predict toy preference, we conducted a meta-analysis of observational studies of the free selection of toys by boys and girls aged between 1 and 8 years. From an initial pool of 1788 papers, 16 studies (787 boys and 813 girls) met our inclusion criteria. We found that boys played with male-typed toys more than girls did (Cohen's d = 1.03, p < 0.0001) and girls played with female-typed toys more than boys did (Cohen's d = -0.91, p < 0.0001). Meta-regression showed no significant effect of presence of an adult, study context, geographical location of the study, publication date, child's age, or the inclusion of gender-neutral toys. However, further analysis of data for boys and girls separately revealed that older boys played more with male-typed toys relative to female-typed toys than did younger boys (ß = 0.68, p < 0.0001). Additionally, an effect of the length of time since study publication was found: girls played more with female-typed toys in earlier studies than in later studies (ß = 0.70, p < 0.0001), whereas boys played more with male-typed toys (ß = 0.46, p < 0.05) in earlier studies than in more recent studies. Boys also played with male-typed toys less when observed in the home than in a laboratory (ß = -0.46, p < 0.05). Findings are discussed in terms of possible contributions of environmental influences and age-related changes in boys' and girls' toy preferences.