Molecular‐equilibrium (chemical‐equilibrium) problems are among the most important, and at the same time most complex and difficult general chemistry problems. In this work, we examine the effect on student performance of the manipulation of the logical structure as well as of the M‐demand of these problems. In addition, we study the relationship between student performance on the problems and a number of cognitive variables, viz., developmental level, working memory capacity, functional M‐capacity and disembedding ability of students. In addition, we test the validity of the Johnstone–El‐Banna model. Nine problems of varying number of operative schemata, as well as of varying number of M‐demand, were used. The number of schemata varied from a minimum of two, to a maximum of four, while within each logical schema we had specific mental demand (M‐demand), varying from four to six. As basic molecular‐equilibrium schemata in these problems, we considered the following: (1) the process of establishment of chemical equilibrium; (2) the condition of chemical equilibrium; (3) the case of gaseous systems, with use of partial and total pressures as well as of Kp; and (4) the disturbance of equilibrium and the establishment of a new equilibrium. The correlation between student performance in the schemata and the steps of the problem solutions is low when the number of schemata is low, but high when this number is high. Of the cognitive factors studied, developmental level plays the most important role, especially as the logical structure of the problem increases. The other three cognitive variables also have an effect, with working memory capacity leading. The findings are mainly attributed to the fact that the prob‐lems were rather algorithmic exercises for the students, because of familiarity and training. © 1998 John Wiley & Sons, Inc. Sci Ed 82:437–454, 1998.