Of the nine South African provinces, Limpopo Province has produced the worst Grade 12 results in the ten years between 1996 and 2006. Yet Thohoyandou and Mutale districts in that province performed outstandingly well from 1994 to 2006. This study aimed to find out what influential factors made the difference within these two districts between high-performing and poorly performing schools, by comparing those that performed well with those that performed badly. Data were gathered by means of questionnaires completed by a total of 87 teachers, and structured interviews were conducted with principals. The sample of 24 schools was divided into two types, ‘high performing’ and ‘poorly performing’. By means of the questionnaires, a total of 114 variables were explored, encompassing a number of different factors, from which a total of 18 hypotheses were derived: three concerning teacher factors (qualifications, home language), 10 concerning school factors (locality, number of classrooms, assembly hall, library etc. and teacher-generated problems), one concerning parental involvement, two concerning teacher motivation and management (frequency and scheduling of staff meetings), and two concerning learner motivation and management (including performance with notes and summaries). The data from both groups of schools were compared in order to test the 18 hypotheses on the influence of different variables upon Grade 12 results, the null hypothesis being, of course, that there was no influence. Before this was done, pairs of variables were also compared and subjected to chi-square testing for each of the two groups of schools to see which of the factors might be related to one another in some way, thus impacting on interpretation of the results of the hypothesis testing. For each of the two groups of schools, results from the questionnaires were analyzed by means of: <ul> <li>Frequency analyses and descriptive statistics extracted from the tables of results where they were of possible interest. Variables such as the gender of teachers were, for example, included.</li> <li>Contingency tables with chi-square analysis testing the independence of the variables where possible relationships between the variables could emerge.</li> <li>Contingency tables with chi-square analysis where the connection was not significant and independence of the variables from each other could therefore be assumed.</li></ul> The chi-square analysis tested the difference between the variables at a 0.5% level of significance. Results of the various analyses were not particularly conclusive. Those with the most reliable levels of significance suggested that the most important variables were those relating to interpersonal relationships, specifically those involving different types of contact. Where staff meetings were scheduled and not particularly frequent, Grade 12 results were better. These results were also better where there was frequent contact between parents and teachers. The results of this study and in this sample area suggest that a school’s management style is more important to Grade 12 performance than the provision of funding and facilities. Further investigation is needed before these conclusions can be generalized to other districts and provinces.