Many epidemiological studies and clinical trials have been performed concerning actinic keratoses. The most eligible endpoint in the majority of articles is counting of actinic keratoses before and after treatments, nevertheless some authors support that this is not a reliable form of evaluation. The aim of this study was to evaluate the actinic keratoses counting by various raters and suggest approaches to increase the reliability. Cross-sectional study: forty-three patients were evaluated by four raters (inter- and intra-rater assessment) on the face and forearms. The mean actinic keratoses counts on the face and forearms were 7.7 and 9.1. The overall agreement among the raters for the facial and forearm actinic keratoses was 0.74 and 0.77. The intra-rater assessment showed high rates of agreement for the face (ICC=0.93) and forearms (ICC=0.83). Higher agreement occurred when counting up to five lesions. Four raters led to increased measurement variability and loss of reliability. Higher rates of agreement may be achieved with small number of lesions, limitation and/or segmentation of body areas to reduce their number, in AK prevention designs, are strategies that may lead to a greater reliability of these measurements.