Predictive models for lag phase duration (lambda) have been less reliable than specific growth rate (mu) models due, in part, to the influence of the pre-growth environment on lambda. A discrete modelling approach was taken to more completely define the response of individual cells to new environments. Time to detection (t(d)) data was obtained from serial twofold dilutions of Listeria monocytogenes growing in a Bioscreen at 30 degreesC. Comparison of the inoculum densities required to achieve maximum t(d) at growth pH values from 7.2 to 4.7 revealed that, as the growth pH decreased, fewer cells were capable of making the transition to the exponential phase. The proportion of these cells (termed "adaptable cells") in the original inoculum was used to define a new parameter (r(0)) which, when combined with the constant mean individual cell physiological state parameter (p(0)), the variation in p(0) (SDp0), the inital inoculum (N-0) and the maximum population density (N-max) was able to simulate a complete growth curve. Power transformations with rescaled explanatory variables provided suitable models for the influence of pH on mu, r(0), and SDp0 (r(2) > 0.70). (C) 2002 Elsevier Science B.V. All rights reserved.