The objective of this study was to calculate the environmental impact of raising pigs fed diets with 192, 182, 172, or 162 g kg^-1 crude protein (CP) from 15 to 30 kg live weight, using a life cycle assessment (LCA) based on either predicted or observed performance and excretion data. The predicted data were calculated for two animal profiles: one with feed intake and weight gain data being representative of the national Brazilian average (y_p-AVG), and the other being representative of the specific experimental farm conditions (y_p-SPC). Performance and excretion of pigs were either measured (y_obs) or predicted (y_p) for each CP scenario by using InraPorc^® model and LCA impacts were calculated from these data. The y_p and y_obs data of performance, nutrient excretion, and LCA were then compared using a Bayesian paired approach. Whatever the calculation methods, the decrease in dietary CP reduced the environmental impacts per kilogram of weight gain on acidification potential (AP), eutrophication potential (EP), and land occupation. Predicted data from y_p-SPC did not differ from y_obs data except for nitrogen excretion and AP and EP impacts. Predicted data from y_p-AVG differed from y_obs for all criteria except for body weight gain, phosphorus excreted, and cumulative energy demand, mainly because of the lower feed conversion ratio for average data. It was concluded that the use of predicted data calculated from specific farm indicators could better explain the environmental impact of pig farming in Brazil compared with the use of national average indicators.