Almost 20 yr ago, the U.S. Nuclear Regulatory Commission (NRC) thermal-hydraulic code development effort made the transition from a homogeneous equilibrium formulation to a two-fluid formulation. The objective was to introduce a more physically based model so that the code analyst would have to make fewer choices in the input deck to simulate expected phenomena. There were still several options left open to the user, especially the noding for the simulation. Recent experience with NRC analyses, as well as with International Standard Problems, has shown that there can still be a considerable “user effect” in the use of even the two-fluid codes. Two specific examples are given. Using the RELAP5 code as a specific prototype to focus the discussion, examples are given of the choices currently available to the analyst. Similar choices are available in almost all thermal-hydraulic system codes. These example choices serve to show that even though thermal-hydraulic system codes are reaching a certain state of maturity, the user must still make many choices in setting up an input deck or in running a calculation. There are several pitfalls that the user can encounter, and there are good practices that can avoid many of these pitfalls. Specific examples of current practices for minimizing pitfalls and increasing good practices are discussed. They apply to any thermal-hydraulic system code. These include the following: training and mentoring for the code analyst; user guidelines documentation; internal review and quality assurance of the input deck by knowledgeable individuals; use of standard noding; and fixed noding for each test facility and reactor system. Guidance for the average code analyst is provided in terms of common pitfalls typically encountered and suggestions for good practices in choosing input deck options.