The performance of the Perez models for global, diffuse and direct luminous efficacy for a Southern Hemisphere site have been evaluated using data collected at Paraparaumu Beach, New Zealand (-41 degrees S, 175 degrees E). All the luminous efficacy models displayed qualitatively the same trends as those measured. Quantitatively, the global luminous efficacy predicted is some 10% lower than measured at the site for all sky conditions. The diffuse luminous efficacy is also 10% lower for intermediate and overcast skies; and accurate (+/-3%) for clear skies. The direct luminous efficacy also showed the same 10% difference for all sky conditions. The discrepancy between measured and modelled data is tentatively attributed to differences in aerosol levels between the rural New Zealand site and the Northern Hemisphere sites used in developing the original models.