Transfer of dietary fatty acids (FA) to milk has been proposed to fit a 2-pool model with a fast pool representing direct transfer by chylomicrons and a slow pool representing indirect transfer through tissue recycling. The objective of this experiment was to quantify direct and indirect transfer of omega-3 (n-3) FA to milk after an abomasal bolus infusion of n-3 FA. Ten ruminally cannulated multiparous Holstein cows (247 ± 113 DIM; mean ± SD) were used in a crossover design with 7 d periods. Cows were milked 4 times daily (6 h intervals) starting 2 d before initiation of the experiment. Treatments were abomasal infusion of 120 g (infused over 1 h) of a free FA mixture enriched in α-linolenic acid (18:3 n-3; EALA) or in very long chain n-3 (>18C n-3; EVLC). The EALA and EVLC treatments provided 80.2 g and 87.6 g of n-3 FA, respectively. Milk was sampled at each milking for determination of milk fat yield and FA profile. The day before bolus infusion was used as a baseline. Total transfer of n-3 FA was analyzed in a model that included random effects of cow nested in sequence, sequence and period and fixed effect of treatment and milk yield (JMP Pro). Time course data was analyzed as repeated measures in SAS and resulting least squares means were fit to a double exponential decay function by nonlinear curve fitting (JMP Pro). Total transfer of n-3 FA to milk differed between treatments (P < 0.001) and was 48.2 and 32.7% of the bolus for EALA and EVLC, respectively. Milk n-3 FA concentration and yield peaked at 12 h and returned to baseline at 138 h post infusion in both treatments. Time course of n-3 FA transferred to milk fit a biexponential model (R² = 0.99). The area (% of total) under the first exponential representing direct transfer was 83.9 and 42.2% and the second exponential representing indirect transfer was 16.1 and 57.8% of the total n-3 FA transferred for EALA and EVLC, respectively. In conclusion, n-3 FA differed greatly in their transfer efficiency mainly due to differences in their direct transfer rates. These differences presumably occur due to trafficking of very long chain n-3 FA into plasma lipid pools unavailable to the mammary gland.