The ability to determine winter frost resistance of woody plants is limited for two reasons: (1) assessment of frost damage in midwinter is extremely difficult because results obtained by the currently available viability assays deviate greatly and (2) equipment that allows plants to be frozen at controlled freezing and thawing rates to below the midwinter frost resistance of most Northern Hemisphere woody plants is unavailable. To overcome these limitations, we developed a novel low-temperature freezing system (LTFS) that makes it possible to conduct in situ freezing experiments in midwinter with full control of cooling and thawing rates down to -70 degrees C. Frost resistance can be determined unequivocally by the regrowth test. The LTFS was tested on various, mostly subalpine, woody plants. Results obtained demonstrate the importance of conducting frost tests in situ. In needles of Picea abies (L.) Karst., frost injuries were not visible immediately after the frost test but took several weeks to develop fully. The low-freezing temperatures attained and the small control oscillations (typically +/-0.1 K) of the LTFS during cooling permitted in situ detection of low-temperature freezing exotherms in xylem of Quercus robur L. and in buds of P. abies and Rhododendron ferrugineum L., all of which showed supercooling. With the LTFS, the effects of low temperatures on plants can be specified directly by in situ assessment and regrowth tests.