We targeted quantitative trait loci (QTL) for milk protein percentage (P%) in two Italian Holstein granddaughter design families using selective genotyping in combination with high throughput amplified fragment length polymorphism (AFLP) markers. A total of 64 extreme high and low sires in respect to estimated breeding value (EBV) for P% (EBVP%) were genotyped with 25 AFLP primer combinations that revealed 305 and 291 polymorphisms in the two families. Association between markers and EBVP% was investigated by a linear model only on bands having paternal origin (105 and 96 AFLP bands in family D and S, respectively). Although no marker was significantly associated with the target trait after correction for multiple comparisons, 17 AFLP markers, significant without correction for multiple tests, were considered suggestive of the presence of a QTL. Eleven of these were successfully located on six Bos taurus (BTA) chromosomes by radiation hybrid or in-silico mapping. Ten of these mapped in the immediate neighbourhood (less than 10 cM) of already described QTL for P%. Suggestive association was verified in four regions by microsatellites analysis: one on BTA 10; one on BTA 28; and two on BTA 18. Microsatellites identified significant effects by single marker and interval mapping analyses on BTA 10 and BTA 28, while they were only suggestive of the presence of QTL on BTA 18. In summary, our results firstly indicate that AFLP markers may be used to seek QTL exploiting a selective genotyping approach in GDD, a wide used experimental design in cattle; secondly, propose two approaches for AFLP mapping, namely in-silico mapping exploiting most updated release from the bovine whole genome sequencing project, and physical mapping exploiting a panel of Bovine/Hamster Radiation Hybrids; and thirdly, provide new information on QTLs for an economic important trait in a never investigated Holstein cattle population. AFLP in combination with selective genotyping can be a useful strategy for QTL searching in minor livestock species, sometimes having large economic impact in marginal areas, where more informative markers are still poorly developed.