This Technical Paper explains the formation of representative households and factor groups for inclusion in the PROVIDE Project Social Accounting Matrix. A general guideline for forming household and/or factor groups is that they should reproduce the socio-economic stratification within the society as accurately as possible. Such groups should also be made up of 'relatively homogenous' groups that are easily recognisable for policy purposes. Here the Income and Expenditure Survey (2000) and Labour Force Survey (September 2000) are used to find demographic statistics, income and expenditure patterns, and employment data used for the creation of representative household and factor groups. The current PROVIDE Social Accounting Matrix groups households according to a range of characteristics, first disaggregating by province and race, and thereafter factors such as agricultural employment, geographical location of the household (homeland or non-homeland), gender and education status of the head of the household, and income level of the household come into play. Provincial factor groups are disaggregated by race and skill level/occupation of the factor. Some alternative household and factor groupings, some of which were used in previous versions of the PROVIDE Social Accounting Matrix, are also discussed, as are a range of other characteristics that may be used to form interesting household groups for future Social Accounting Matrices.