In this work, three-phase liquid-phase microextraction (LPME) based on a supported liquid membrane (SLM) sustained in the wall of a hollow fiber was investigated with special focus on optimization of the experimental procedures in terms of recovery and repeatability. Recovery data for doxepin, amitriptyline, clomipramine, and mianserin were in the range of 67.8–79.8%. Within-day repeatability data for the four basic drugs were in the range of 4.1–7.7%. No single factor was found to be responsible for these variations, and the variability was caused by several factors related to the LPME extractions as well as to the final HPLC determination. Although the volume of the SLM varied within 0.4–3.1% RSD depending on the preparation procedure, and the volume of the acceptor solution varied within 4.8% RSD, both recoveries and repeatability were found to be relative insensitive to these variations. Thus, the handling of microliters of liquid in LPME was not a very critical factor, and the preparation of the SLM was accomplished in several different ways with comparable performance. Reuse of hollow fibers was found to suffer from matrix effects due to built-up of analytes in the SLM, whereas washing of the hollow fibers in acetone was beneficial in terms of recovery, especially for the extraction of the most hydrophobic substances. Several of the organic solvents used in the literature as SLM suffered from poor long-term stability, but silicone oil AR 20 (polyphenyl-methylsiloxane), 2-nitrophenyl octyl ether (NPOE), and dodecyl acetate (DDA) all extracted with unaltered performance even after 60 days of storage at room temperature.