The angular size-redshift test for quasars was compared with various cosmological models including non standard models. The possible effects of radio source orientation and relativistic beaming were taken into account in the analysis. It was found that orientation effects alone were not sufficient to explain the observedθ-z relation in terms of Friedmann models. In addition, linear size evolution of the formD ~ (1 +z)−n , with 0.75 ≲n ≲ 1.2 would be required for 0 ≲ Ω ≲ 1.0, or possibly an inverse correlation between luminosity and linear size. The non-standard cosmological models all gave better fits to the deprojected data than the Friedmann models in the absence of evolutionary effects, with the tired light effect providing the best fit.