Out of 30 infrared pasteurization tests, those providing more than 4-log reduction in Pediococcus population size on the kernels and involving short holding times were selected for producing samples for sensory analysis (Table 2). The sensory attributes were evaluated by a panel of 80 untrained panelists. Each panelist was given five cups of almonds each of which contained five almond kernels; one was labeled as reference and contained raw almonds and the other four contained randomly selected IR-pasteurized almonds, based on the conditions listed in Table 2, unknown to the panelists. A difference test was carried out in which the panelists were asked to compare the reference with each of the treated samples in terms of appearance, texture, flavor and overall quality. The panelists were asked to compare the reference with the IR-pasteurized almonds and to select ‘‘Yes’’ if they perceived a difference in the above-mentioned attributes and ‘‘No’’ otherwise. The answer of each panelist for each attribute was classified as successful when the answer was Yes, and as failure when the answer was No. The probability of success for each question was 0.5. Since each sensory attribute was evaluated by at least 40 panelists, the binomial distribution approximated to a normal distribution. Therefore, the data of each treatment were analyzed by the binomial distribution test using the BINOMDIST built-in function (assuming a noncumulative distribution) in MS Excel (Microsoft Co. Ltd., Redmond, WA, USA) at a significance level of a = 0.01.