We argue that the conventional approach to bankruptcy modelling, which relies on accrual-based ratios, is vulnerable to the earnings management of a company threatened by insolvency. This fact may pose significant limits on the possibilities of distress prediction. Business distress is defined as cashflow insufficiency, and cashflow indicators are less vulnerable to earnings management. For these reasons we assume that cashflow ratios are theoretically more suitable for predicting distress. In our research we analysed the usefulness of cashflow-based ratios as potential predictors of bankruptcy. During the research, the cashflow-based ratios take the form of many variants of operating, financial, investment and free cashflow into a firm in combination with total assets, sales, liabilities and other indicators. The research was carried out on a sample of 4,350 Czech manufacturing SMEs operating during the period between 2013 and 2018. We employ the previously published approach of hybrid modelling to create the prediction model, though we propose a modification for the purposes of this paper. The modified hybrid model employs Classification and Regression Trees and Logistic Regression, while we use the Principal Component Analysis method to deal with the problem of multicollinearity. The results showed that operating cashflow ratios play a significant role in financial distress, especially when combined with short-term debts.