Recent critiques in the stress literature have identified several methodological shortcomings, including ignoring control over event occurrence, the confounding of life event and outcome measures, the lack of consistent and reliable measures, and few cross-validational studies. Each of these problems is addressed empirically by examining two independent samples of adolescents. A 39-item life event measure, developed previously in one sample of 1018 adolescents, was administered to a second sample of 376 teenagers. Second-order factors of six stress scales were clustered into controllable and uncontrollable events, both of which were related to psychological distress. Poor health was significantly related to an uncontaminated factor of uncontrollable events. Means and covariances of the seven stress factors were compared between the derivation sample and an age-matched subset of the cross-validation sample. Correlates of the scales with drug use and personality revealed interpretable associations within the cross-validation sample and were quite similar to those in the original sample. Finally, few sex differences were evident, whereas scores on controllable events scales increased with age.