The concept of long-lived magma chambers in which goechemical evolution is controlled by periodic replenishment, tapping (eruption), and crystal fractionation, the so-called RTF chamber, is the subject of previous numerical models by O'Hara (1977) and O'Hara & Mathews (1981). These authors explore the properties of idealized steady-state chambers, using constant parameters to express amounts of replenishment, fractionation and tapping. The present work introduces randomizing factors into these parameters, to attempt to simulate the natural world more closely. Comparison is made with continental flood basalt (CFB) sequences which erupt in a quasi-steady state, the geochemical characteristics of which can readily be matched by numerical models, although no detailed inversion of the natural data is possible. However, if such sequences do in fact originate in RTF chambers, conventional estimates of the mass of concealed cumulates will generally be under-estimates. Additionally, it is possible to place limits on the magnitudes of the x (amount of fractionation) and y (amount of tapping) parameters, and on the x/y ratio, the main factors controlling general levels of trace element concentrations. Small values of x and y (e.g., 0.1 or less) coupled with large values of x/y (e.g., 10 or more) can theoretically produce dramatic geochemical effects, but although the CFB data are consistent with the former, the x/y ratio appears to be constrained to the range 0.5–1.3. Changes in inter-incompatible element ratios observed in a Deccan traps sequence studied in detail cannot be interpreted as RTF-generated in a system with constant-composition magma input and no assimilation or contamination.