One of the main shortcomings of standard sampling synthesis is the very limited number of sound parameters that are user-controllable. In the most general case, the user can choose a particular pitch, duration, and amplitude. If the sampler allows control over articulation, it simply switches from one sound sample to another. This makes fine-tuning of musical performances demanding and time-consuming if not an impossibility altogether. A synthesis system has been developed at the Academy of Music in Krakow, Poland. It uses a large collection of samples that contain short sequences of notes. The system implements a number of techniques to seamlessly connect recorded sequences, to control note durations as well as the tempo and the dynamics envelopes. Samples are automatically chosen, modified, and connected to keep the recorded, natural note transitions intact. The system uses performance rules to introduce variations into the regular playback akin to live performances by musicians. A user can either control the parameters manually or choose a desired expression and leave the particular decisions to the system. However, it is necessary to examine which parameters have the greatest impact on the listeners’ impression and determine useful values. 15 expert listeners compared and evaluated variants of musical performances produced by the synthesis system with different sets of parameters. The paper discusses a selection of the examined parameters, the test methods employed and the results obtained.