Prior literature studying railway accounting during the nineteenth and twentieth centuries defends the thesis of lack of reliability of accounting figures. This prior research, which mainly studies the cases of the United Kingdom and the United States, offers mixed views on the causes, or simply accepts this thesis without providing conclusive evidence, as is the case of historical research in Spain. We provide novel evidence on the quality of railway accounting and contribute to this prior debate by (1) analysing the accounting for two material accruals: depreciation and prior period adjustments; (2) studying the persistence of earnings and its components, and (3) analysing how accrual accounting affects persistence. These analyses are conducted for the period 1856-1939 for the two major Spanish railway companies (MZA and NORTE). The reported evidence suggests that earnings are highly persistent. However, we show that there are significant differences across firms and that these differences are particularly obvious when analysing the adjustments for prior period earnings. Overall, our evidence does not support the thesis that accounting was underdeveloped, but rather, that managerial accounting choices lowered accounting quality.