The scintigraphic detection of sentinel lymph nodes (SNs) in early-stage breast cancer is a widely accepted diagnostic method. However, which radiotracer administration mode should be used is still controversial. This prospective study aimed to intra-individually compare the detection rates obtained after peritumoural versus subareolar injection with regard to SN number and localisation. Fifty-one women (age, 32-76 years) with breast cancer were investigated on two consecutive days. On day 1, 140-400 MBq technetium-99m nanocolloid was injected along the peripheral tumour margins. Static lymphoscintigrams of the axilla, thorax and neck were taken in various views 1 and 19 h p.i. On day 2, 10 MBq (99m)Tc-nanocolloid was injected subareolarly in the clock position of the tumour and dynamic and static scans were performed immediately. Thereafter, 30 MBq (99m)Tc-nanocolloid was administered peri-subareolarly and lymphoscintigrams were acquired in a dynamic and static manner. In 49/51 women, the different injection techniques disclosed the identical number and location of SNs in the axilla. In seven patients, the peritumoural injection detected additional SNs in the parasternal group. Axillary SNs were detected as early as 2-15 min following subareolar injection, both in the clock position and peri-subareolarly, as compared with about 1 h after peritumoural administration. Sixteen patients showed at least one tumour-positive SN, and nine also had tumour-positive non-SNs. One patient with a tumour-negative SN, visualised concordantly by both subareolar and peritumoural administration, demonstrated two metastatic non-SNs, yielding a false-negative rate of 5.9%. In conclusion, a simple subareolar injection in the clock position is sufficient for SN detection in breast cancer, if it is accepted that parasternal lymph node detection has no therapeutic consequences.