Fisheries management is difficult especially in developing countries where there are little or no data available for stock assessment. Here, a simple model based on the length-based Bayesian biomass (LBB) is applied to length frequencies collected on the two sardinella species (Sardinella maderensis and S. aurita) collected in Senegalese waters (2004 to 2014) to diagnose these stocks and to support the development of fishery management options that may improve the livelihoods of artisanal fishermen. Annual mean length of both species of sardinella showed a large variation during the decadal study period. It is assumed that such variations are due to environmental changes. According to our results, based on the current exploitation rate, both sardinella species are overexploited. To reverse these bad stock status, three management indicators were estimated for both sardinella: (i) length at first capture (Total Length (TL); 24 and 27 cm for S. maderensis and S. aurita, respectively); (ii) the length at maximum possible yield per recruit (TL; 26 and 29 cm for S. maderensis and S. aurita, respectively); and (iii) the optimal length for the first capture (TL; ; 25 and 28 cm for S. maderensis and S. aurita, respectively). According to Senegalese maritime fishing code, which sets the small pelagic fish size of the first capture 18 cm, we urge the Senegalese governments to press ahead with much needed reforms of the fishing code. We recommend capturing sardinella at the size to provide a natural safeguard against any recruitment failure related to environmental variability and allow individuals to grow and ensure the long-term survival of populations and thus sustainable fisheries. The results suggest that LBB model could be a tool to assess data-poor fisheries allowing the possibility to include in the analysis several years of length–frequency data with a minimum of prerequisites.