While there is now a considerable body of research in Australia and elsewhere that has revealed clear evidence of beta instability, only a small proportion of the resulting papers examine the important associated issues of risk estimation and modeling; and asset pricing tests which incorporate time-varying beta. This paper has conducted an exploratory ex post investigation of these issues using a sample of monthly data on Australian industry portfolios over the nineteen-year period 1974 to 1992. Specifically, industry betas were modeled, estimated and tested with reasonable success in terms of: (1) regimes related to periods of regulation/deregulation/imputation; (2) the level of market returns; and (3) a measure of volatility on the risk-free rate of interest. However, univariate and multivariate tests reported in the paper provide mixed evidence concerning the applicability of a time-varying beta GAPM, that incorporates these variables. The analysis of this paper has been primarily illustrative in nature. It leaves open the way for an extended investigation of the many issues identified throughout which remain unresolved.