Purpose. To test the predictive ability of loan asset indicators on Commercial bank fragility in Kenya. 
 Design/Method/Research approach. The study adopted positivism research philosophy with exploratory research design. The study population was 42 Commercial banks in operation on 31st December 2015. Secondary data was collected from Central Bank of Kenya and analysed using Stata Statistics/Data analysis. Generalised Linear Model was used to establish the relationship between asset indicators and bank fragility. The concept of credit creation was explored as the genesis of bank fragility. This study is part of early warning systems in detecting bank fragility.
 Findings. The research found a direct relationship between a lagged dependent variable, loan portfolio growth, loan deposit ratio and bank fragility.
 Practical implications. Recommendations are followed on the basis of this study. At first, regulator develop a potential solution to control loan portfolio growth, cap loan deposit ratio and limit the level of non-performing loans. Banking practitioners should model monthly reporting requirements to ensure that banks are able to disclose the ratio and explain any significant changes. Secondly, since Non-performing loans can act as an incentive for bank managers to seek deposits and lend more thereby exacerbating the problem, banks with NPL to gross loans greater than an upper threshold determined by the regulator should not be allowed to attract more deposits. Thirdly, set the maximum level of loan deposit ratio to avoid expensive, sensitive and high-risk loan capital. Implementation of these recommendations will lead to secured social welfare.
 Originality/Value. The study examines the role of certain loan asset indicators on bank fragility and extends the discussion in the area of early warning systems and commercial bank instability in Kenya.
 Research limitations/Future Research. This research contributes to the discussion on bank fragility and early warning systems. The further research should review evidence from other jurisdiction with high numbers of distressed institutions to determine how many months or years before distress the three significant variables could predict fragility. Besides, there is need for research on insider loans as defined and why there was no statistical significance.
 Paper type. Empirical.