Maldistribution of local anesthetic administered through a subarachnoid catheter recently has been implicated as a possible cause of sacral root injury. To examine subarachnoid distribution of catheter-injected local anesthetic, we constructed a model of the subarachnoid space and administered solutions containing lidocaine and methylene blue through sacrally directed catheters. We studied three catheters: a 28-G endport, a 20-G endport, and a 20-G multiple sideport. To determine the injection rates to be used, ten clinicians were observed while they performed mock subarachnoid injections: the mean (+/- standard deviation) "normal" injection times for the 28-G and 20-G catheters were 52.6 +/- 17.2 and 11.9 +/- 7.2 s, respectively. The correlation coefficient for lidocaine concentration estimated by methylene blue spectrophotometric absorbance and measured by immunoassay was 0.977. Administration of hyperbaric local anesthetic through a sacrally directed catheter resulted in restricted distribution of anesthetic with a relatively high peak concentration. Rate of injection was a critical factor affecting distribution; faster injections tended to distribute solution more uniformly and to a higher segmental level, resulting in substantially lower peak concentrations. When catheters were injected at clinically relevant rates, the 28-G catheter produced the greatest degree of maldistribution; this difference appeared to be primarily a function of flow rate. Differences in peak lidocaine concentration between the two 20-G catheters were neither large nor consistent. However, despite sacral placement, the multiple-sideport catheter distributed anesthetic toward "higher" spinal segments more consistently. Distribution was more favorable when the injected solution was less dense (closer to isobaric).(ABSTRACT TRUNCATED AT 250 WORDS)