Purpose This paper provides evidence that the banking sector reforms of 2004 and 2009 enhanced prudential performance of the banking industry and financial system stability in Nigeria. Design/methodology/approach This study uses regression analysis with regime shift to confirm results from tests of two means and variances model to examine the effectiveness of banking sector reforms in Nigeria. Findings Evidence from the regression model agrees with findings from the test of means model (not controlling for trend effects) that capital to assets ratio rose while non-performing loan ratio declined after the reforms, and that capital to earning assets ratio rose when trend effects were accounted for. Both the regression model and the tests of means model controlling for trend effects show that return on asset, return on equity and return on earning assets ratios declined after the reforms. Research limitations/implications This paper evaluated the effectiveness of banking sector reforms in Nigeria using models that avoid weaknesses that besieged many previous studies. It however used data covering 1983–2020 period, due to data availability. A larger scope of data may improve the results, and future research may re-examine this theme as more data become available. Furthermore, banking stability issues could be examined using specialised techniques such as the generalised autoregressive conditional heteroscedasticity model and related family. Practical implications These results suggest that the reforms led to improvement in the sector’s resilience (risks-absorbing capacity) and asset quality, and that profitability had not been the primary focus of the reforms. Social implications The authors recommend that regulatory and supervisory authorities in Nigeria continue to implement and improve on banking sector reforms for a more resilient and functional banking system. As a contribution to social research, this study shows that studies on policy evaluation should be located within appropriate theoretical framework: the theory of change. It shows that an appropriate use of attribution analysis and contribution analysis within this theoretical framework engenders robust analysis and results. Otherwise, the analytical findings would be erroneous and policy advice misguided. Originality/value The statistical significance of our findings establishes that the banking sector reforms in Nigeria have been effective in promoting financial system stability in Nigeria. By deploying both the test of means with and without trend effects (an attribution analysis) and the multivariate regression analysis with regulatory shift (a contribution analysis), and relying more on the later for its superiority, this study contributes to the body of knowledge in that, it not only determined the true effects of banking sector reforms in Nigeria for appropriate policy guidance but also demonstrated that, in research, an inappropriate methodology produces results that may diverge from the more accurate ones that were derived from the correct methodology.