The experimental Hot Dry Rock (HDR) geothermal reservoir created in 1985 at the Rosemanowes test site in Cornwall is one of a very small number of HDR reservoirs of substantial dimensions whose key performance parameters have been measured. The reservoir linked an injection well (RH12) with a production well (RH15). The minimum separation of the openholes was 133 m. The initial production temperature (at the casing shoe) was 80-degrees-C. The average injection temperature during the early part of the circulation test was about 25-degrees-C. Ten per cent thermal drawdown occurred after about 300 days＇ circulation at an average production flow rate of 8.5 kg/s. The mass of fluid produced in this time was 220 million kg. At the lowest injection pressure of about 4 MPa, the production flow rate was 4 kg/s, the impedance was 1.0 MPa/(kg/s), and the water loss was 1 kg/s (i.e. 20% of the injection flow rate). At higher injection pressures, impedance was reduced but water loss was increased. A comprehensive programme of reservoir characterisation was undertaken to measure as many as possible of the fundamental reservoir parameters that are considered to have given rise to the observed thermal and hydraulic performance. This paper reports the main results of this reservoir characterisation programme. It is concluded that many of the parameters measured are not true estimates of the actual parameters being sought, but apparent parameters, whose values are highly dependent upon the simplifying assumptions used to obtain them. The performance of the RH12/RH15 reservoir is compared with that which would be required of an individual cell of a commercial-scale HDR reservoir consisting of many stimulated cells in parallel. By these criteria, the performance of the RH12/RH15 reservoir falls short in terms of thermal output and water loss, but meets the required impedance. The performance of the RH12/RH15 reservoir was severely limited by the unfavourable trajectory of the injection well (RH12), especially in terms of thermal output. The results from this experimental reservoir are therefore equivocal as far as validation of the HDR doublet concept is concerned.