This paper presents a new multivariate approach to spatial autocorrelation analysis of genetic data, applicable to multiallelic codominant, multilocus arrays. The method treats genetic data as a whole, reducing stochastic noise and enhancing spatial signal. It uses genetic distance methods and nonparametric permutational testing for full correlograms. The approach is illustrated with data from the orchid Caladenia tentaculata, showing that multiallele treatment provides clearer results than single-allele analysis. Intermediate frequency alleles from highly polymorphic loci perform well, while rare alleles do not. Multilocus analysis provides clearer answers than single-locus analysis, and differential weighting of alleles improves resolution minimally. The method is more general than previous approaches and can be applied to various genetic markers, including SSRs, RAPDs, and AFLPs. The paper also discusses the relationship of this method to other spatial autocorrelation techniques and highlights the importance of multivariate analysis in understanding spatial genetic structure. The results, though specific to Caladenia, suggest potential for broader application in plant population genetics.This paper presents a new multivariate approach to spatial autocorrelation analysis of genetic data, applicable to multiallelic codominant, multilocus arrays. The method treats genetic data as a whole, reducing stochastic noise and enhancing spatial signal. It uses genetic distance methods and nonparametric permutational testing for full correlograms. The approach is illustrated with data from the orchid Caladenia tentaculata, showing that multiallele treatment provides clearer results than single-allele analysis. Intermediate frequency alleles from highly polymorphic loci perform well, while rare alleles do not. Multilocus analysis provides clearer answers than single-locus analysis, and differential weighting of alleles improves resolution minimally. The method is more general than previous approaches and can be applied to various genetic markers, including SSRs, RAPDs, and AFLPs. The paper also discusses the relationship of this method to other spatial autocorrelation techniques and highlights the importance of multivariate analysis in understanding spatial genetic structure. The results, though specific to Caladenia, suggest potential for broader application in plant population genetics.