A meta-analysis of available data on dose response to dietary phosphorus (P) in fish from over 70 feeding trials reported in 64 published studies covering over 40 species of fish was performed. Broken-line regression was used to model the data sets. The meta-analysis showed that estimated minimal dietary P level varies with the response criterion and that estimates should preferably be expressed in terms of available P than in terms of total P. Estimates based on whole-body P concentration (4.7 g available P kg−1 dry matter, DM) or vertebral P (5.2 g available P kg−1 DM) were greater than that for maximizing somatic weight gain (WG) (3.5 g available P kg−1 DM) or plasma P concentration (2.8 g available P kg−1 DM). P content of fish varies linearly with body mass (3.6 g kg−1 live weight). Use of ingredients rich in P or of diets with high basal P content or high levels of water P concentration can affect the estimations. Among the different response criteria tested, WG was found to be the most reliable and whole-body P concentration to be the most stringent criterion to estimate P requirement of a given fish species. Expressing available P requirement as g P per unit DM or digestible energy (DE) in the diet was equally effective, but expressing in terms of g P intake per kg BW0.8 per day would be more precise.