Background: Longitudinal research on determinants of initiated and continued benzodiazepine (BZD) use is inconsistent and has identified many possible determinants. It is unclear which of those are most important in the prediction of BZD use. We aimed to identify the most important predictors of initiated and continued BZD use. Therefore, we analyzed the most consistently identified determinants from previous research plus some new determinants. Methods: We identified baseline and 2-year longitudinal predictors of initiated BZD use (vs nonuse) among 2205 baseline BZD nonusers and of continued use (vs discontinued use) among 369 baseline BZD users in the Netherlands Study of Depression and Anxiety using logistic regression analyses. Results: During follow-up, BZD use was initiated by 4.9% of BZD nonusers at baseline. Initiated use was predicted by insomnia (odds ratio [OR], 1.60), enduring anxiety symptoms (OR, 2.02), entering secondary care during follow-up (OR, 2.85), and past BZD use (OR, 3.57). Positive life events during follow-up reduced the likelihood of BZD initiation (OR, 0.76). Of BZD users at baseline, 54.2% continued use during the entire follow-up period. Continuation of BZD use was predicted by higher age (OR, 1.03), severe anxiety (OR, 1.85), and a long duration of BZD use (OR, 1.54). Leaving secondary care was associated with less continued BZD use (OR, 0.29). CONCLUSION:: Insomnia and anxiety were the main risk factors of initiated use, whereas advanced age and anxiety severity were the main risk factors of continued use. Sex, education, pain, and physical health seemed to be less important. © 2011 by Lippincott Williams & Wilkins.