A 2-year field study on kennelled stray dogs living in a highly endemic area of leishmaniasis was designed to evaluate whether deltamethrin-impregnated collars (Scalibor) Protector Band) could confer protection against leishmaniasis in this peculiar setting, and to assess differences in clinical outcomes between collared and uncollared dogs. A cohort of 120 clinically healthy and Leishmania-seronegative dogs was enrolled, 50% of which were collared before the 2003 transmission season, and then re-collared before the subsequent season. Collared and uncollared animals were allowed to live with infected dogs in same groups within the kennel. Follow-up included serological (IFAT) assessment twice a year with parasitological Leishmania confirmation, and clinical evaluation performed every 3 months on seroconverted dogs from both groups. Collar losses during the two seasons were high (35%). About 50% of enrolled dogs were lost at follow-up because of death or they were moved to other locations. After the 2003 season, cross-sectional serological examinations tested positive in 5 out of 44 collared animals (11.4%) and in 14 out of 34 controls (41.2%), with 72.3% estimated protection (P<0.005). After the 2004 season, 7/31 seronegative collared dogs seroconverted (22.6%) compared with 7/17 seronegative controls (41.2%), with 45.1% protection (P=0.15). At the end of the study, the cumulative rate of protection was 50.8% (P=0.005). At the clinical evaluation of 21 seroconverted dogs from both groups, canine leishmaniasis signs were significantly more frequent (90% versus 36%, P=0.017) and rapidly progressive in uncollared than in collared dogs. Reasons for such partial clinical protection in collared dogs may be found in the vector anti-feeding effect of protector bands, resulting in a lower number of infectious bites and, probably, in the reduction of antigenic stimuli necessary to shift toward a non-protective immune response.