A free choice type experiment was made, with two treatments (line IAC 73/228 and cultivar Paraná) and thirty replications, inside a screen cage with artificial infestation of Nezara viridula (L.) (Hemiptera : Pentatomidae). The level of infestation used in this experiment was, approximately, 1 adult per plant during the reproductive phase of the plant. There was no significant difference in the number of adults infesting both treatments. The number of damaged pods was also not significantly different in both treatments. However, the number of sound pods was 3.6 times grater in the line IAC 73/228, since this line had a total number of pods of 291 per plant, in average, whereas the cultivar Paraná had 100, and out of this total, the line IAC 73/228 had, in average, 248 sound pods per plant whereas the cultivar Paraná had only 68.8. The mean yield obtained in the screen cage infested by the insect, was compared with the one obtained in an uninfested screen cage set beside it: under infestation, the line IAC 73/228 yielded about 7 times more than Paraná and in the absence of infestation, yielded twice as much. The cultivar Paraná had a decrease in yield of 67.5%, due to the infestation, whereas the line IAC 73/228 had an increase in yield of 3.4%. The line IAC 73/228 showed a good level of tolerance to Nezara viridula, and it was more productive than the cultivar Paraná even in the absence of infestation under the experimental conditions of screen cage, late plating and wide spacing between plants. Pod samples taken at the top and middle part of the plants showed greater damage at the top. The values of the F test for treatments were greater for samples taken from the middle part of the plants. A damage index for pods was calculated as: IDP = % of intermediate pods + 2 (% of flattened pods). The values of the F test between treatments using the IDP, for both regions, the top and the middle part of plants, were greater than the F values obtained using the percentages of sound or damage pods. A similar damage index for the grains has also been proposed by GILMAN et alii, and is calculated with the formula: IDG = 0 (% of sound grains) + 1 (% of grains with light damage + 2 (% of grains with moderate shrunking) + 3 (% of grains with severe shrunking). The values of the F test between treatments were, however, greater for the percentage of sound grains than for the IDG. All the plants of the cultivar Paraná remained green at harvest time, showing foliar retention called in Brazil "crazy soybean". Only 14.8% of the plants of the IAC 73/228 exhibited this sympton. Based on the above results, the authors recommended for criteria for soybean selection for resistance to stink bugs, under conditions of severe infestation: 1: the index of pod damage obtained from a sample taken from the middle part of the plants; 2: the percentage of sound grains obtained from the same sample; 3: the percentage of plants with foliar retention at harvest time; and 4: yield.