Fish recruitment in riverine reservoirs is not fully understood because the long-term data series requiredfor standard stock–recruitment models are often lacking. In this study, two unrelated piscivorous specieswith different ecologies, asp (Leuciscus aspius) and pikeperch (Sander lucioperca), were investigated over a14-year period in a reservoir in the Czech Republic using a novel informative statistical approach based ondimension reduction methods. This method is useful for situations in which potential predictors are equalto, or exceed, the length of the time series. Recruitment of asp fry was affected by zooplankton abundance,predator density and temperature. Recruitment of pikeperch fry measured with seine and trawls was onlyaffected by the number of predators, while recruitment of pikeperch fry estimated with gillnet data wasalso affected by temperature and water level fluctuation. Although gillnets are commonly used samplingmethod, it seems to be inappropriate for developing fry predicting model. This research also highlightsthe use of a novel approach to dimension reduction for analysis of factors affecting recruitment usingshorter time series (in our case 13 years).