In recent years several investigators have studied the self-diffusion coefficient of zirconium, Dv, both in the alpha phase of the pure metal and in dilute zirconium alloys [1][2][3][4][5][6][7][8][9]. In fact, the first results were published by research groups of the Soviet Union [1][2][3][4] and Argentina [7] and, in both cases, the measurements were performed at temperatures near the ~ to/~ phase transition and in polycrystals.Hood and Schultz [8] measured Dv in e-Zr single crystals at 1124 K, oriented for diffusion parallel and perpendicular to the hexagonal axis. The most recent and important work in this field, however, was presented by Horvfith et al. [9] where the self-diffusion coefficient was measured between 779 and 1128K, using ion-beam-sputtering techniques. The authors found an anomalous behaviour in the metal, because Dv does not follow an Arrhenius law. No physical model has been given by the authors to explain this anomaly. Hood [10] has suggested that such anomaly might be produced by the more rapid diffusion of impurities present in e-Zr but, to date, no clear explanation has been given for this behaviour.It is the purpose of this letter to introduce new elements into the discussion of the self-diffusion coefficient of e-Zr, including new values obtained from an analysis of results obtained by mechanical testing, in the temperature region 633 to 695 K.The results found by Horv~ith et al. [9] indicate that Dv is, approximately, two orders of magnitude lower than the value reported by Dyment and Libanati [7], which was attributed to a higher dislocation density, causing a higher diffusion rate through the dislocations core. This interpretation agrees with the theoretical work of Hart It is interesting to point out that the anomalous variation of Dv with the temperature can be accounted for by considering an analogous electrical model of fluxes in series [15], where