Abstract Data from 30 studies with 173 treatment mean observations of lactating goats were used to determine metabolizable protein (MP) requirements for lactation (MPl). Milk protein yield (MkP) was calculated from milk yield and protein concentration. MP was estimated from dietary ingredient composition and a feedstuff database of CP degradability properties and ruminal fermentable energy concentration. MPl was estimated with a factorial method by subtracting MP used for maintenance, which was the sum of scurf (0.2 g/kg BW0.6), endogenous urinary (1.03 g/kg BW0.75) and metabolic fecal CP losses (2.67% DM intake) divided by an efficiency of use of MP for maintenance of 1.0. Also, MP was adjusted for BW change (14.3% protein), assuming an efficiency of MP use for protein accretion of 0.59 and that mobilized tissue protein was used for lactation with the same efficiency as MP from the diet or microbial cells. The equation for the regression of MPl (g/day) against MkP (g/day) was: MP l =15.2 ( S.E. =7.77)+1.30 ( S.E. =0.090)× MkP (n=163, adjusted -R 2 =0.56) ; the intercept was not different from zero (P>0.05). The slope of a no-intercept equation (regression line forced through the origin) was 1.45 (S.E.=0.033). In conclusion, these results suggest a MPl requirement for goats of 1.45 g/g of MkP or a milk protein efficiency of 0.69. Because of the approach employed, application of this MPl requirement should not include provision of additional MP as a safety factor. With the large number of observations in this database, this requirement estimate should be of value in expressing protein needs and predicting performance of lactating goats, although future research to refine assumptions may improve accuracy.