An attitude questionnaire, administered to third, fourth, and fifth grade students, is discussed. The questionnaire, "How I Feel," was composed of a packet of directions to the teacher, directions to the student with sample questions, and 11 concepts, each of which was evaluated by six bipolar adjective pairs that were chosen because of their relevance to the type of information desired by the researchers and by their high loading on Osgood's evaluative dimension. Nine of the concepts were scored whereas two served as validity-type scales to identify those students who did not understand what they were asked to do or who may have filled in the blanks with random checks. The students to whom the questionnaire was administered were 238 who were attending a school using Individually Prescribed Instruction (IPI). Each of the individual bipolar adjective pair scales was scored along a five-point continuum. The questionnaire was administered again to 201 of the original subjects seven months following the first administration, using the same instructions but a different "special" teacher. Results showed the coefficient alpha reliabilities for the individual scales ranged from .86 to .92. A factor analysis of the entire instrument indicated that the six bipolar adjective pairs used to evaluate the nine concepts were essentially measuring the same factor and could therefore be summed, thus reducing the original 54-items to 9 items. The test-retest reliabilities for six of the nine questions ranged from .35 to .44; they were considerably lower for the other three questions. Tables present the data, and appendixes provide the questionnaire and related materials. (DB)