Purpose: Retrospective evaluation of the accuracy of manual multisegmental irradiation with a source train for irradiation of long (re)stenotic lesions in coronary arteries, following percutaneous transluminal coronary angioplasty (PTCA). Material and methods: Thirty-six patients were treated with intracoronary irradiation following PTCA with manual multisegmental irradiation. These patients were included in the multicenter, multinational ‘European Surveillance Registry with the Novoste Beta-Cath system’ (RENO). In all 36 patients the target length (i.e. PTCA length plus 5-mm margin at each side) was too long for the available source train lengths (30 and 40 mm). In 33 patients the radiation delivery catheter was manually positioned twice and in three patients three times in series, trying to avoid any gap or overlap. The total number of junctions was 39. Following a successful PTCA procedure the site of angioplasty was irradiated using the Novoste Beta-Cath afterloader with a 5-F non-centered catheter which accommodates the sealed beta-emitting 90Sr/ 90Y source train or dummy source train. Radiation was delivered first to the distal part of the target length. Fluoroscopic images of this source position were stored in the computer memory. For irradiation of the proximal part of the target length, the delivery catheter had to be retracted over a distance equal to the source length used for the distal part. This was done by a continuous overlay video loop with ECG-gated replay of the image stored in the computer memory. The dummy source was used to position the delivery catheter so that the junction between both source positions was as precise as possible. Measurements of gap or overlap between the source positions were performed retrospectively on printed images. Doses were calculated, in accordance with the Novoste study protocol, at a distance of 2 mm from the source axis (=dose prescription distance) in several points along the irradiated length. Results: Interventional or PTCA length varied between 33 and 95 mm. The lesion sites were in the left anterior descending artery, ( n=6), right coronary artery ( n=20), left circumflex artery ( n=6) and one vein graft. The administered radiation dose was determined by the vessel diameter and the presence of a stent. This dose, prescribed at a distance of 2 mm from the source axis, varied between 16 and 22 Gy. No gap or overlap was seen between the two source trains in only two out of 39 cases. In 16 cases there was a gap ranging between 0.6 and 9.6 mm and 18 cases showed an overlap of 0.5–14.4 mm. In three patients the measurement was not possible. In case of a gap the minimal dose calculated at 2 mm from the source axis varies between 0 and 87% of the prescribed dose, depending on the distance between both sources. In case of overlap the maximal dose varies between 110 and 200% of the prescribed dose at 2 mm from the source axis. Conclusions: The results show the inaccuracy of manual multisegmental irradiation using a source train in coronary arteries, causing unacceptable dose inhomogeneities at a distance of 2 mm from the source axis at the junction between both source positions. Moreover, a perfect junction will never be possible due to movement of the non-centered radiation delivery catheter in the vessel lumen, as applied in this study. Manual multisegmental irradiation is therefore not recommended. Using longer line sources or source trains or preferably an automated stepping source is a more reliable and safer technique for treatment of long lesions.