In modern swimming pool research reactors the stable isotope 9 Be from the neutron reflector made of metallic beryllium is converted into 10 Be via a (n, γ )-reaction. This isotope contaminates the primary cooling water as a corrosive of the reflector material and has to be monitored continually. Since 10 Be can only be determined with nuclear spectrometric methods after an extensive radiochemical procedure, an indirect two step procedure was worked out: ( a ) calculation of the specific 10 Be-activity from the accumulated radiation periods since installation of the reflector into the reactor and ( b ) determination of the total amount of beryllium in the cooling water by using a sufficiently sensitive analytical technique. Knowledge of both values allows the immediate calculation of the 10 Be contamination of the pool water. It is shown that the inductively coupled plasma optical emission spectrometry (ICP-OES) coupled with ultrasonic nebulization is applicable to the determination of beryllium in water samples in the sub-ppb region. Comparing the three different emission lines, 234.861 nm, 313.042 nm and 313.107 nm, the first one proved to be best suited for the above mentioned analytical problem. The Be determination in the reactor water samples was also carried out using electrothermal atomic absorption spectrometry (ETAAS) with longitudinal heating of the graphite tube and Zeeman background compensation. The samples were measured without matrix-modification using the 234.861 nm absorption line. Quality assurance for both analytical methods was done determining two reference water samples. Detection limits (3σ) were similarly low: 0.012 µg l -1 for ICP-OES and 0.005 µg l -1 for ETAAS. Parallel analysis of seven different reactor pool water samples using both methods showed concentration values in fairly good agreement. The concentrations obtained were in the range 0.02–0.46 µg l -1 and yielded 10 Be activities of 0.3–9 Bq l -1 .