The use of observational data, collected during the routine clinical care of patients, has been advocated as a means to obtain clinically relevant information regarding the pharmacokinetic parameters of drugs. However, the validity of this approach and its proper role in new drug development is unclear. This study was performed to evaluate the ability of three approaches to estimate population pharmacokinetic parameters: the traditional approach, mixed-effect modeling, and a simple pharmacokinetic screen. The evaluation was performed with data collected during a multicenter, open-label study evaluating the efficacy, safety, and pharmacokinetics of imipramine and alprazolam in combination. The traditional pharmacokinetic study demonstrated a 20% decrease in the clearance of imipramine in the presence of 4 mg/day alprazolam. Mixed-effect modeling extends these findings by suggesting that the interaction is dependent on the simultaneous concentration of alprazolam, a finding that was not possible under the study design typically used for traditional pharmacokinetic studies. Although the simple screen suggests the presence of the drug-drug interaction, limited information regarding pharmacokinetic parameters is available and those parameters that can be estimated are biased.