Contrary to common perceptions, the reliability of quantitative assays for the determination of drugs in biological fluids using high-performance liquid chromatography with tandem mass spectrometric (LC/MS/MS) detection methods and the integrity of resulting pharmacokinetic data may not be absolute. Results may be adversely affected by lack of specificity and selectivity due to ion suppression caused by the sample matrix, interferences from metabolites, and "cross-talk" effects. In this paper, an example of the effect of the sample matrix on the determination of finasteride (I) in human plasma is presented. The ion suppression effect was studied by analyzing standards of I injected directly in mobile phase and comparing the response (peak areas) of I and an internal standard (II) with the peak areas of the same analytes spiked before extraction into five different plasma pools and standards spiked into the plasma extracts after extraction. The LC/MS/MS analyses were performed using a turbo ion spray interface (TISP) under chromatographic conditions, characterized by minimal (total run time of 2 min, capacity factors, k' of 1.50 and 1.75 for I and II, respectively) and high retention of the analytes (total run time 6 min, k' of 3.25 and 13.25 for I and II, respectively). The absolute peak areas for I and II in different plasmas were calculated, and the slopes and peak area ratios at all concentrations within the standard curve ranges were compared. When analyses were performed under conditions of minimal HPLC retention, the slope of the standard line for one set of plasma samples was substantially different (about 50% higher) from that from other plasma sources. The precision of the assay, expressed as coefficient of variation (CV, %) was also inadequate and varied from 15 to 30% at all concentrations within the standard curve range. When the same experiments were repeated using high HPLC retention, the slopes from different plasma sources were practically the same, and the CV was improved to 6-14%. By increasing k' and providing more chromatographic retention of analytes, the "unseen" interferences from plasma matrix were mostly separated from analytes, practically eliminating the ion suppression. In addition, by eliminating from plasma extracts a number of endogenous components through more selective extraction, the ion suppression was also minimized. The detailed data and the design of these experiments are presented. In addition, development of a highly sensitive assay for I in human plasma at low picogram per milliliter concentrations using LC/MS/MS with a heated nebulizer (HN) interface, instead of a TISP interface, is described. In this case, the effects of sample matrixes were not observed.