Temperature variations during high pressure processing (HPP) can induce uncertainty in process characterization, particularly for determination of microbial inactivation kinetics. Ultra-high temperature (UHT) whole milk inoculated with Staphylococcus aureus ATCC 12600 or Escherichia coli O157:H7 ATCC 43895 was exposed to 600 MPa at 4, 21 and 45 degrees C. An isothermal-endpoint HPP procedure led to constant process temperature during pressure increase and so minimized temperature change involved in HPP. Three nonlinear models: the Weibull, log-logistic and modified Gompertz methods, were fitted to survival data. Goodness-of-fit of these models were compared using mean square error. Overall, S. aureus and E. coli O15TH7 showed less inactivation at 4 degrees C than at 21 and 45 degrees C. The log-logistic model was a better fit for E. coli O157:H7 inactivation data. In contrast, either the modified Gompertz or the Weibull models, depending on the process temperature, were more appropriate to describe S. aureus inactivation. (c) 2005 Elsevier Ltd. All rights reserved.