The effects of students' grade level (four, five, or six) and of five variables related to types of open arithmetical sentences on students' ability to solve open-sentence problems were investigated. The five variables were: (1) operation symbol, (2) sentence type as determined by the symmetric property of the equality relation, (3) the position of the placeholder, (4) whether the solution was a whole number or not, and (5) whether the largest number in the problem occurs as a product in a basic fact. Two instruments incorporating these variables were developed and administered to 1298 students in grades four through six. Responses were coded and analyzed using the Fortap Statistical Package. Data from students whose performance meets predetermined minimal students were further analyzed using ANOVA and Wilcoxen signed rank tests. These analyses indicated that sentences with whole number solutions were significantly easier than others. Variables (1), (2), and (3) were also related to significant differences in subject performance, and several significant interaction effects were found. (SD)