Smallholder farms in humid tropics often display complex associations of perennial and annual crops as farmers tend to valorise between-row spaces of the perennials by intercropping with shorter cycle crops, all in a bid to better absorb climatic and associated economic risks. To better understand the functioning and performance of such systems, the flow of farm inputs and production should be properly monitored. This study was conducted to assess the suitability of traditional land evaluation and multivariate statistical analysis to define relatively homogeneous entities or farm section units (FSU). Data obtained from several perennial crop-based farms were analysed using a traditional land evaluation technique and multivariate statistical analysis. Irrespective of the evaluation method, all parameters varied considerably for the plots surveyed. The zones differed with respect to their altitudes, farm holding size, annual rainfall, and slope of fields. Although both evaluation systems seemed complementary, results obtained did not always converge owing to differences in the nature and magnitude of criteria considered. These results demonstrate the feasibility of complementing traditional land evaluation with some multi-variate statistical analysis to classify perennial crop based farming systems.