The ability of Snyder's theory of linear gradient elution to predict the starting isocratic reversed-phase LC conditions (k' = 4-10) for the opioid peptides was investigated. The errors in predicting the concentration of acetonitrile (phi) required to elute the peptides with a k' value of 4 were high, ranging from 13.5 to 38.1%. At k' = 10 the errors were generally reduced to less than 20%. This analysis was repeated with the same peptides after conversion to their fluorescent 1-cyanobenz[f]isoindoles (CBIs) by reaction with naphthalene-2,3-dicarboxaldehyde/cyanide. For the CBI derivatives, the errors in predicting the required concentration of acetonitrile for isocratic elution were markedly reduced and ranged from 0 to 14.3 for k' = 4 and 0 to 11.9% for k' = 10. The errors in the model in predicting the required isocratic mobile phase accurately were attributed to a mixed mechanism of retention involving solvophobic and silanophilic interactions and leading to non-linear relationships between log k' and phi. Even when the errors in predicting the required value of phi were relatively high, the Snyder approach was found to be very useful in predicting the initial starting conditions for the reversed-phase LC of the native opioid peptides as well as their fluorescence CBI derivatives.