We examined the diagnostic efficiency of symptoms for experimentally diagnosed oppositional defiant disorder (OD) and attention-deficit hyperactivity disorder (ADHD), and for their differential diagnosis, in a nonreferred sample. More specifically, we used four conditional probability indices--positive predictive power, negative predictive power, sensitivity, and specificity--in an attempt to evaluate the diagnostic efficiency of these symptoms. Teacher ratings of OD and ADHD symptoms were gathered on 102 boys, ages 8 to 12. On average, ADHD symptoms were as useful as OD symptoms as exclusion criteria for OD, while OD symptoms were nearly as effective as ADHD symptoms as inclusion criteria for ADHD. Nonetheless, a number of OD and ADHD symptoms appeared to be useful as both inclusion and exclusion criteria for their respective disorders. These findings illustrate the utility of conditional probability indices in both the diagnosis and differentiation of childhood disorders.