Detection thresholds are typically obtained by presenting a subject with serial dilutions of an odorant. Many factors, including the solvent used to dilute the odorant, can influence the measurement of detection thresholds. Differences have been reported in detection thresholds for phenyl ethyl alcohol (PEA) when different solvents are used. In this study we used gas chromatography (GC) to investigate further the effect of solvent on odor detection thresholds. We used a single ascending method and serial dilutions of PEA in four different solvents--liquid paraffin (LP), mineral oil (MO), propylene glycol (PG) and dipropylene glycol (DPG)--to determine the PEA thresholds for 31 adult subjects. For each solvent, we prepared eight serial log base 10 step dilutions (1-8), with corresponding liquid PEA concentrations of 6.3 x 10(1)-6.3 x 10(-6) (% v/v). We found that the threshold concentrations for PEA in LP (step 6.5) and PEA in MO (step 5.5) were significantly lower (P < 0.05) than for PEA in PG (step 4.0) and DPG (step 4.0) We then used GC to measure both the liquid and gas PEA concentrations for the dilution steps prepared with LP and PG. Although there were large threshold differences in the liquid concentrations of PEA in LP and PG, the headspace gas concentrations of PEA were the same. These results demonstrate the importance of determining the gas concentration of odorant stimuli when performing odor threshold measurements, in particular when comparing odor detection thresholds obtained using different solvents.