Instrumental and sample matrix factors were examined for their effect on the resolution, sensitivity and detectability of oligodeoxynucleotides separated on 50-mu-m I.D. gel-filled capillaries. Substantial errors in the calculated percentage purity may result from using peak areas not corrected for peak velocity. Area multiplied times velocity, not area, is used as the quantitative response with units of area counts cm/min. At high oligonucleotide concentrations the response varies linearly with increasing sample oligonucleotide concentration, but at low concentrations the response varies anomalously. This finding indicates the importance of using internal standards for quantitations. Increases in the sample conductivity dramatically decrease the sensitivity, whereas changes in the sample pH have only a minor effect. The mass loading increases linearly with increase in the injection voltage and duration. With injection durations of up to 20 s, the sensitivity is increased in excess of one order of magnitude with only a 15% decrease in resolution. The separation efficiency of larger sized oligonucleotides decreases more readily with increasing on-column mass than smaller species. Increases in the separation field strength and column length increase the resolution consistent with theory only when the detector rise time is reduced. An increase in the column temperature results in a decrease in the migration times by 1.1% per degrees-C with only minor decreases in resolution.