AbstractForty‐nine boys, aged 3 to 11 years, who were referred for an evaluation of a potential gender identity disorder of childhood, were assessed with a standardized in vivo behavioral observation of sex‐typed play in an alone condition and with the parent‐report Bates‐Bentler Games Inventory. Without access to these measures of sex‐typed play, an independent clinical psychologist rated each of the subjects on a 9‐point scale that indicated the degree of severity of a gender disturbance, based upon child and parent interviews and a complete psychological test battery. The percentage of feminine play in the behavioral observation procedure was found to be correlated significantly with the clinician rating of degree of gender disturbance (r=.55). The Games Inventory was found to differentiate significantly the gender‐disturbed boys from the nonclinical reference group on its feminine/preschool games, masculine nonathletic games, and athletic games subscales and on its composite index of feminine play preference; however, only the composite index (and none of the 3 subscales) was correlated significantly (r=.34) with degree of severity of gende