The grafted human glioblastoma cell CB109 was used as a model for intralesional therapy with 131I-labelled hyaluronectin glycoprotein (131I-HN).131I-HN bound specifically to in situ hyaluronic acid (HA), a main component of the extracellular matrix which is involved in tumour invasion. Labelling experimental conditions were determined and, finally, 25 microCi/microgHN, 1 microg chloramine-T/microgHN and a 60-s stirring period provided a 131I-HN preparation with an optimal affinity for HA (64% compared to unlabelled HN). Following intratumoral injection, 131I-HN was retained with a limited diffusion outside the tumour. On day 4 the radioactivity concentrated in the tumour was still 25 times greater than that in the liver, spleen and kidneys combined. For therapeutic assays, 65 microCi 131I-HN was injected into the tumour, resulting in a delivery of 6.8 Gy over a 7-day period. Controls received unlabelled HN, heat-inactivated HN, a mixture of inactivated HN plus free 131I or no treatment (six animals per group). Tumour volumes were evaluated every second day from treatment day and the rate of tumour growth was expressed as a ratio of tumour size at time intervals to the tumour size at the time of injection. Growth curves were compared: heat-inactivated with or without free 131I had no anti-tumour effect. Unlabelled HN-injected tumours had a slightly slower growth rate than untreated tumours (p < 0.02) and growth rate of 131I-HN-injected tumours was much lower (p < 0.00002). A pronounced inhibitory effect with intralesional 131I-labelled HN injection resulted from a combination of a) blockage of HA, a proliferation facilitating factor, and b) local irradiation of tumoral tissue, while uptake in normal tissues was minimized.