Fixation with formalin, a widely adopted procedure to preserve tissue samples, leads to extensive degradation of nucleic acids and thereby compromises procedures like microarray-based gene expression profiling. We hypothesized that RNA fragmentation is caused by activation of RNAses during the interval between formalin penetration and tissue fixation. To prevent RNAse activation, a series of tissue samples were kept under-vacuum at 4°C until fixation and then fixed at 4°C, for 24 hours, in formalin followed by 4 hours in ethanol 95%. This cold-fixation (CF) procedure preserved DNA and RNA, so that RNA segments up to 660 bp were efficiently amplified. Histological and immunohistochemical features were fully comparable with those of standard fixation. Microarray-based gene expression profiles were comparable with those obtained on matched frozen samples for probes hybridizing within 700 bases from the reverse transcription start site. In conclusion, CF preserves tissues and nucleic acids, enabling reliable gene expression profiling of fixed tissues.