This paper classifies households for the purposes of rural food security analysis using the multivariate statistical technique of factor analysis. This technique, applied to the social sciences, assumes that socio-economic behaviour can be related to stable underlying structures. These could be stable social strata dictating particular tendencies. With appropriate variables, factor analysis can reveal underlying social structures, and in the process identify important variables accounting for variation. The technique is preferable to univariate analysis, because it relies less on prior subjective assessment of the importance of explanatory variables. However, the major weakness of factor analysis is that it assumes the existence of well developed stable structures with unique distinguishable characteristics. Complications arise if the variables under analysis are subject to structural change. In this case, some subjectivity is inevitable since unambiguous factor classifications become implausible. Despite these problems the technique produces interesting and useful results for the peasantry in Zimbabwe. For several years during the first decade of independence, Zimbabwe boasted aggregate food surpluses-- to which the peasantry contributed significantly. This 'success story', however, masks a different reality, one in which many rural inhabitants remained poor and food insecure, and in which an even larger number of families was highly vulnerable to poverty. Based on information collected for 67 households in two communal areas of Mashonaland West, we demonstrate that household food security is a serious problem in Zimbabwe. Our analysis shows that only a small group of households are responsible for the production successes. Households classified as 'poor' and 'vulnerable', together constitute more than three quarters of the sample population. This is more clearly apparent when production, marketing and income data are examined over a two year period (1986/87 - 1987/88), including one drought year. Some members of the 'vulnerable middle' stratum also do well in terms of production, output sales and food security, but only in years of adequate rainfall. In the drought year, the economic condition of the vulnerable households is less robust, support from relatives, and from the state, is needed to avert disaster. An interesting finding, but not uncommon to Southern Africa, is that household survival across all the strata is highly dependent on access to non-farm incomes. These are especially important for improving the food security of 'poor' households. An implication of the findings of factor analysis is that food security policies in Zimbabwe need to pay more attention to addressing a complex set of inter- and intra-household distributional problems. Attempting to enhance rural household food security by primarily raising aggregate food output, be that for national, regional, or local markets, is likely to prove inadequate. A number of inequalities need to be recognised within the peasantry. For example, in the distribution of productive resources, in the distribution of well-paid rural and urban jobs, in the gender distribution of labour effort, and in the distribution of incomes within families. As shown in the paper, the peasantry in Zimbabwe is divided along many lines, and failure to take that into account, in terms of food and agricultural policy, can often lead to misplaced emphasis and action. Factor analysis can help to distinguish economic strata in the countryside and identify some of the basic causes of household food insecurity within different strata. A first step towards improving food security is to know the different characteristics of insecurity and poverty. The paper contains a preliminary check-list of indicators to help policy makers and development practitioners identify poor and vulnerable households in regions 11 and ifi of Zimbabwe. The authors are aware that several other dimensions affecting food insecurity in communal areas are not dealt with in this paper. The examination of other important influences on individual and household food security, will be the subject of subsequent papers, e.g., the effect of different marketing structures, of gender differences within and across households, and of different cropping patterns.