A dinâmica parasitária da Haematobia irritans foi estudada em três fazendas com diferentes tipos de manejo de bovinos mestiços na região de São Carlos, SP, as quais não utilizavam nenhum tratamento específico à mosca. O experimento consistuiu na contagem de mosca na região dorso-lombar a cada 14 dias no período de outubro de 1992 a outubro de 1994. Cada fazenda adotava um tratamento: T1 (Fazenda Ribeirão Bonito, verminose tratada com levamizole-Ripercol L injetável, carrapato com piretróide - Ectoplus "pouron" e berne com trichlorphon-Neguvon + óleo queimado, usotópico); T2(Fazenda São Carlos, verminose tratada com levamizole - Ripercol injetável, carrapato tratado com banho de imersão com formamidina - Triatox e berne com fenthion -Tiguvon "pour on") e T3 (Fazenda Santa Eudóxia - sem tratamento). Os dados de contagem da mosca-do-chifre (MC), na escala √(MC + 0,5), foram analisados em um modelo que incluiu além da média, os efeitos de tratamentos, ano, mês, período e as interações duplas, sendo que todos os efeitos diferiram entre si (PHaematobia irritans parasitical dynamics was studied in three farms rearing crossbred beef cattle, under different management systems, in the region of São Carlos, São Paulo State. In the farms was not utilized any specific treatment against horn flies. The experiment consisted in counting fortnightly the number of flies on the dorso-lombar area of the animals, from October, 1992 to October, 1994. Each farm used a different parasite control: T1- Ribeirão Bonito Farm, anthihelminthic treatment, with injectable levamizole (Ripercol L); ticks treated with pour on pyrethroid compound (Ectoplus); and berne, treated with trichlorphon (Neguvon) plus used motor oil, applied topically; T2- São Carlos Farm, antihelminthic treatmen with injectable levamizole (Ripercol L); ticks treated with formamidine (Triatox) in immersion bath; and berne treated with pour on fenthion (Tiguvon); and T3- Santa Eudóxia Farm, no parasitism treatments. The number of horn flies (MC) was analised in the scale √(MC + 0,5) in a statistical model that included, besides the mean, the effects of treatments, year, month, period and the double interactions. In the scale MC, all the effects differed from each other (P < 0.01). During the months in which the temperature was higher (October to March) average parasite load was of 15.7±36.4 flies, while temperature was lower (April to August) it was of 6.8 ± 23.6 flies. The average parasite load differed (P < 0.01) among years (92 = 4.8 ± 7.8; 93 = 8.8 ± 20.0; 94 = 16.6 ± 43.8) and among farms (T1= 3.1 ± 9.7; T2= 2.2 ± 13.7 e T3=26.0 ± 44.5), indicating that horn fly behaviour is irregular and noncyclic. The efficiency of 81.1% and 86.3% was observed for treatment T1 and T2, respectivelly, when compared to treatment T3.