A previously published method was applied to the prediction of chilling times for hoki (Macruronus novaezelandiae). The method was selected for its computational simplicity and applicability to objects of irregular shape. The method was tested against experimental data from trials in which individual hoki were chilled in flaked ice or an ice/water slurry. Analysis showed that heat transfer along the length of the hoki body made only a small contribution to temperature change at the thermal centre, and that hoki can therefore be modelled as an infinite ellipse to estimate the shape-dependent model parameters E and L. Predicted and measured values of E showed satisfactory agreement, but there was a systematic offset between predicted and measured values of L, leading to overprediction of measured chilling times by an average of 17.4 +/- 2.6%. Differences between predicted and measured chilling times were highly variable. However, both offset and variability were largely attributable to experimental and data uncertainties. The method for predicting chilling time is recommended as a useful tool for the analysis and design of chilling processes for hoki. (c) 2005 Elsevier Ltd. All rights reserved.