When electrodermal activity (EDA) recordings are controlled for artifacts, i.e., electrodermal reactions [EDRs] elicited by breathing irregularities, several problems arise. For example, respiration is difficult to evaluate because there are no clear-cut criteria for its values, e.g., wave form, depth. Furthermore, respiration and EDA are rather complexly intertwined, and there is no established or standardized method for evaluation. Especially when subjects are not stimulated, i.e., when nonspecific EDRs are taken, EDR recordings elicited by irregular breathing may overestimate the subject's arousal and bias any given research question. Moreover, incidences of concurrent consecutive EDRs and changes in respiratory activity may encourage multicausal interpretation due to both signals' having a common central causation. To circumvent such problems, we developed a method which provides rule-based guidelines to identify potential artifacts. Two experiments (N = 14 and N = 12) were conducted to test the accuracy of the judgments of three independent raters. The reliability coefficients for the number of electrodermal reactions and the sum of their amplitudes yielded satisfactory coefficients of convergence for each individual experiment (.87 and .82 in Exp. 1 vs .94 and .95 in Exp. 2) as well as for the two experiments combined (.92 and .91).