The authors of the present study addressed the measurement of temperament by examining the convergence between observational and questionnaire measures and the occurrence of contrast effects in parental ratings of non-twin siblings on the Children's Behavior Questionnaire. Fathers', mothers', and observers' ratings of 94 early-school-age sibling pairs were obtained longitudinally such that siblings were rated at the same time and when they were the same ages. Convergence between laboratory observations and questionnaires was generally modest to moderate for shyness and activity level; low to modest for positive emotionality, fear, and interest; and low for anger and sadness. We found some evidence of contrast effects for activity level and fathers' ratings of shyness and of assimilation effects for anger and sadness. The results highlight the complexity of measuring temperament and suggest that rater biases should be evaluated according to the measurement instrument used.