The amounts of soil phosphate extracted by Truog's, Morgan's, Bray's Nos 1 and 2, 0·3N HCl, North Carolina, Olsen's, 0.01M CaCl2, Egner et al., 0·02N Na2EDTA and anion exchange resin methods of determining available phosphate were compared with the amounts of inorganic phosphate in different forms in ten Bangladesh soils. Correlation studies between the soil test values obtained by any two methods revealed that the correlation coeffi-cient was usually higher when both test values were well correlated with a common phosphate fraction. This indicated a similarity of methods in selective dissolution of specific phosphate forms. Based on significant positive correlations between methods as well as their similarity in selectivity of dissolution of Ca-P, Al-P and Fe-P, the methods were classified into three groups. Truog, North Carolina and 0·3N HCl methods were grouped together, their extractants being capable or bringing Ca-P more effectively into solution and Al-P into solution satisfactorily. Bray's No. 1, Bray's No. 2 and 0·02N Na2EDTA methods formed a second group as their extractants were found to dissolve Al-P preferably and Ca-P moderately. The third group comprised the anion exchange resin and 0·01M CaCl2 methods which were not selective in dissolution or any phosphate form. Methods other than Olsen's and anion exchange resin gave values that were non-significantly and negatively correlated with reductant soluble iron phosphate and occluded iron and aluminium phosphates. Practical implications of the findings were discussed.