Two studies were designed to evaluate the relative bioavailability of l-carnitine delivered by different methods in dairy cattle. In experiment 1, 4 Holstein heifers were used in a split-plot design to compare ruminally or abomasally infused l-carnitine. The study included 2 main-plot periods, with infusion routes allocated in a crossover design. Within main-plot periods, each of 3 subplot periods consisted of 4-d infusions separated with 4-d rest periods. Subplot treatments were infusion of 1, 3, and 6 g of l-carnitine/d in conjunction with 6 g/d of arabinogalactan given in consideration of eventual product manufacturing. Doses increased within a period to minimize carryover risk. Treatments were solubilized in 4 L of water and delivered in two 10-h infusions daily. Blood was collected before the start of infusion period and on d 4 of each infusion period to obtain baseline and treatment l-carnitine concentrations. There was a dose × route interaction and route effect for increases in plasma carnitine above baseline, with increases above baseline being greater across all dose levels when infused abomasally compared with ruminally. Results demonstrated superior relative bioavailability of l-carnitine when ruminal exposure was physically bypassed. In experiment 2, 56 lactating Holstein cows (143 ± 72 d in milk) were used in 2 cohorts in randomized complete block designs (blocked by parity and milk production) to evaluate 2 rumen-protected products compared with crystalline l-carnitine. Treatments were (1) control, (2) 3 g/d of crystalline l-carnitine (crystalline), (3) 6 g/d of crystalline, (4) 5 g/d of 40COAT (40% coating, 60% l-carnitine), (5) 10 g/d of 40COAT, (6) 7.5 g/d of 60COAT (60% coating, 40% l-carnitine), and (7) 15 g/d of 60COAT. Treatments were top-dressed to diets twice daily. Each cohort used 14-d and included a 6-d baseline measurement period with the final 2 d used for data and sample collection, and an 8-d treatment period with the final 2 d used for data and sample collection. Plasma, urine, and milk samples were analyzed for l-carnitine. Crystalline and 40COAT linearly increased plasma l-carnitine, and 60COAT tended to linearly increase plasma l-carnitine. Total excretion (milk + urine) of l-carnitine averaged 1.52 ± 0.04 g/d in controls, increased linearly with crystalline and 40COAT, and increased quadratically with 60COAT. Crystalline increased plasma l-carnitine and l-carnitine excretion more than 40COAT and 60COAT. In conclusion, preventing ruminal degradation of l-carnitine increased delivery of bioavailable carnitine to cattle, but effective ruminal protection and postruminal bioavailability is challenging.