The external heat transfer coefficient in steam retort processing was determined experimentally in a pilot scale retort. The heat transfer equations were solved applying finite elements and using the actual retort temperature profile as boundary condition. The instantaneous values of the heat transfer coefficient were determined, to analyse its time-variability along a retort cycle. It was found that reliable results for the external heat transfer coefficient at time t could be obtained from the derivatives of the average heat transfer coefficient calculated between time zero and time t. The results showed a sharp increase of the hear transfer coefficient in the earlier times of heating (up to 4-5 min of processing), followed by a slightly increasing pattern during the remaining heating period. Using average heat transfer coefficients for the heating phase (and for the cooling phase) also resulted in quite accurate estimates of the temperature at the geometric centre of a can. The influence of the two average heat transfer coefficients, heating and cooling, on the lethality was studied by building a response surface. It was concluded that the variability of the coefficient during heating has a greater impact, especially in the range 150-260 W/m(2) K. (C) 2000