The oxidation of long-chain fatty acids is carnitine-dependent; these fatty acids penetrate the mitochondrion to be oxidized only when they are bound to carnitine in the form of acyl-carnitines. To meet the need for carnitine, animals depend on both exogenous supplies and endogenous synthesis. The aim of the present paper is to review our knowledge of endogenous carnitine synthesis. The precursors of carnitine are lysine and methionine but its true point of origin is trimethyllysine. This molecule is either obtained from the diet or is synthesized in the body from L-lysine (bound to protein) which is methylated 3 consecutive times by an S-adenosyl-methionine. Trimethyllysine is transformed into hydroxy-trimethyllysine, then into trimethylaminobutyraldehyde and finally into trimethylaminobutrate (or gamma-butyrobetaine). The gamma-butyrobetaine is hydroxylated into carnitine. This reaction chain only functions well when three vitamins--ascorbic acid, pyridoxin and niacin--are present. Studies on rat have shown that skeletal muscle, heart, intestines, testis, and especially kidneys, insure the transformation of trimethyllysine into gamma-butyrobetaine but that only the testis, and especially liver, can hydroxylate gamma-butyrobetaine into carnitine. However, in rat the relative importance of the kidneys and liver in total carnitine synthesis has not yet been determined. The situation is the same in man, although it has been proven that human brain and kidneys, as liver, have gamma-butyrobetaine hydroxylase. It is known that the rate of carnitine synthesis depends on three factors--the amount of trimethyllysine available, the rate of gamma-butyrobetaine transfer to tissue(s) hydroxylating it and gamma-butyrobetaine hydroxylase activity. Moreover, it appears that carnitine synthesis is not slowed down by prolonged fasting, that it does not completely cover body needs during the first postnatal days, and that it does not decrease in two patients with systematic carnitine deficiency.