An intercomparison study of the determinations of glyphosate, chlormequat and mepiquat residues in cereals was performed. Four samples comprising one blank, two incurred and one spiked sample were sent to six participating laboratories. For glyphosate, two laboratories reported considerably lower results than the other four. One of the two laboratories with low results also reported low recoveries. The results of a sample spiked with 0.80 mg kg(-1) glyphosate and an incurred sample, ranged from 0.23-0.87 mg kg(-1) and 0.11-0.25 mg kg(-1) respectively. The strong correlation between the two samples (r(2) = 0.95) indicates a systematic between-laboratory variation. Several different principles were used for the analysis of glyphosate using different clean-up techniques and GC/MS, HPLC-fluorescence or LC/MS for detection. The results of the chlormequat residues showed more consistency. All but one laboratory obtained comparable results. However the correlation between the results for the sample spiked with 0.38 mg kg(-1) (range: 0.26-0.65 mg kg(-1)) and the incurred samples (range: 0.19-0.45 and 0.15-0.23 mg kg(-1), respectively) again showed a strong correlation (r(2) = 0.99 and 0.88) indicating a systematic component. For mepiquat, results above the limit of quantification were only reported for the spiked sample. The results ranged from 0.29-0.92 mg kg(-1) (spiked concentration= 0.38 mg kg(-1)). Three laboratories had results that deviated less than 25% from the fortified concentration. Two laboratories reported results 38% and 141% above the fortified concentration, respectively.