This thesis empirically examines whether regularities in UK stock returns can be explained by the Fama and French (1996) three-factor model and whether the factors adopted by this model are truly associated with risks captured by return covariance. It offers valuable independent out-of-sample evidence to safeguard against the data-snooping bias of contemporary literature in this area that is mainly focused on the US. To begin with, I show that contrary to Fama and French (1996), their three-factor model does not explain many existing cross-sectional return regularities based on company-specific characteristics. This result is robust to alternating methodologies similar to Fama and French (1996) and Brennan et al. (1998). The empirical observation that some company-specific characteristics predict return beyond the control of the Fama-French model implies either that there exist non-risk influences in stock returns or that the model itself is misspecified. To assess the latter, I apply the Daniel and Titman (1997, 1998) approach to compare the return predictability of characteristics and covariance pertaining to the Fama-French model factors. I find convincing evidence that stock returns have a stronger relationship with characteristics than with loadings on the factors suggested by Fama and French (1993, 1996). The results we observe consistently reject the risk explanation but not its characteristics counterpart. The results are robust to alternating methodologies applied to address the issues suggested by Berk (2000). The findings in this thesis have implications for various applications such as test of market efficiency, risk management, cost of capital estimations, and performance measurement.