Purine analogues are effective treatment for CLL, but despite improved response rates and durations for purine analogues as compared to chlorambucil no survival advantage was documented. Our trial is the first comparing CdA and fludarabine (F) as first-line treatment of symptomatic B-CLL with high dose intermittent chlorambucil (Chl) as control. Patients aged 18–75 years from Scandinavia, Australia and UK were since November 1997 randomised between monthly courses of Chl 10mg/sqm orally day 1–10, F 25mg/sqm iv day 1–5, or CdA 5mg/sqm iv/sc day 1–5. Protocol is seen at www.cck.ki.se/CLL-study/index.html. Randomisation was stratified according to age and stage. Up to six courses were given according to response and toxicity. We now report a scheduled interim analysis of the first included 150 patients; 139 were evaluable and 5 excluded for not fulfilling eligibility criteria and 6 for incomplete data reporting. There was no significant difference in age, Binet stage or time from diagnosis to inclusion between the groups. Median age was 63 years and median time from diagnosis to inclusion was 16 months. Overall responses according to NCI criteria and intention to treat were documented in 57%/67%/74% of patients in the Chl/F/CdA arms, respectively (Chl vs CdA, p=0.06, Chi-test). The number of CR/nPR/PR in the different treatment groups were for Chl 2/3/22 (n=47), for F 0/2/28 (n=45), and for CdA 2/3/30 (n=47). NCI grade III and IV platelet toxicity was seen in 25% and 20% for Chl, 24% and 5% for F, and 36% and 11% for CdA. NCI grade III–V infections were seen in 25%/28%/30%, respectively. One F patient died in septic shock from E coli, and two CdA patients died from RS virus infection and aspergillosis, respectively; none of them responded to CLL-treatment. Autoimmune hemolysis was seen only with purine analogues, 4 during F and 3 during CdA treatment. We conclude that toxicity was feasible, and infections did not differ between arms. So far there is no significant difference in response rates between the therapies, but CdA seems at least as good as F. Inclusion is currently ongoing but will close during 2004, and a full evaluation is scheduled thereafter.