OBJECTIVES: To evaluate the use of the population-based Emilia-Romagna region (RER) administrative database for drug outcomes studies using as an example the previously documented increased mortality risk amongst new users of conventional versus atypical antipsychotic medications. METHODS: We conducted a new user cohort study among 23,681 Italian RER patients aged 65 or older who initiated treatment with a conventional or atypical antipsychotic between July 1, 2009 and June 30, 2011.  The 180-day mortality was compared for patients in each treatment group using Cox proportional hazards models that were adjusted for mortality risk factors, including demographic and clinical characteristics, use of other medications, and measures of health services utilization intensity, all measured prior to antipsychotic initiation.  We conducted a systematic review and meta-analysis of studies with similar methods against which we compared our results.  RESULTS: Amongst 14,462 patients prescribed conventional and 9,212 prescribed atypical antipsychotics, we observed 2,402 (16.6%) and 821(8.9%) deaths during follow-up, respectively.  New users of conventional antipsychotics were older and generally had higher prevalences of outcome risk factors and higher health service use intensity at baseline.  The crude hazard ratio (HR) was 1.95 (95% confidence interval [CI], 1.80-2.11), which decreased to 1.47 (95% CI, 1.35-1.60) after full adjustment.  We identified seven published studies examining this association using similar methods.  The pooled HR from these seven studies was 1.34 (95% CI, 1.28-1.39).  Upon inclusion of our study, the meta-analysis yielded a summary estimate of 1.35 (95% CI, 1.31-1.40) without introducing any heterogeneity (I2 = 0%; p=0.455). CONCLUSIONS: Our results support the findings of previous studies and provide a refined estimate of the association between increased mortality amongst elderly patients taking conventional versus atypical antipsychotics.  Our study provides support for the use of the RER database for pharmacoepidemiology studies.