Latent growth curve methodology was applied to the study of patterns of change in client resistance during parent training therapy. The clinical sample consisted of 68 mothers of children, 52 boys and 16 girls, ages 5 through 12 years, with confirmed conduct problems. Simple linear and linear spline growth models were fit to the three repeated observational measures of maternal resistance during therapy and found inadequate. Instead, a quadratic growth model was used. Pretherapy maternal characteristics of inept discipline and antisocial behavior predicted chronically high levels of resistance. Maternal depressed mood predicted less negative quadratic curvature. No significant predictors of overall change in resistance were detected. Lack of negative curvature of the resistance growth curves predicted child court offenses during a 2-year posttermination follow-up. Results are discussed with reference to the struggle-and-working–through hypothesis of client resistance.