Although the supercritical-pressure or high-performance light water reactor (HPLWR) concept is largely based on the well-established technological experience available with conventional light water reactors, there is still no consensus on various key design features such as an optimal layout for the fuel assembly. This results mainly from the very large density variations of supercritical-pressure water in the core, which render it difficult to ensure reliable values for parameters such as power peaking factors and reactivity worths. The present paper describes studies carried out to compare deterministic and Monte Carlo codes for analysing a representative square HPLWR lattice with uniform 5%-enriched UO2 fuel. The main purpose has been to assess the prediction accuracies achievable for integral parameters such as the multiplication factor, control absorber effectiveness, moderator/coolant density reactivity feedback and pin power distributions. The results show good agreement between the deterministic and stochastic calculations for the unperturbed lattice. However, for certain perturbed situations involving, for example, local coolant density changes in the assembly or control absorber insertion, the observed discrepancies are large enough to question the basic viability of the reactor physics design, e.g. with respect to the thermal performance. [All rights reserved Elsevier]