Mycoplasma genitalium is an established cause of sexually transmitted infections. Studies of disease associations are often performed on archived specimens, but little is known about the effect of storage of specimens on the detection of M. genitalium. Genital swab and first-void urine specimens submitted for detection of M. genitalium were tested on the day of receipt. Remnants of positive original specimens as well as DNA preparations were stored at -20°C for up to 18 months. A total of 361 M. genitalium-positive specimens were available. PCR after repeat DNA preparation was performed for 262 specimens. The sensitivity after repeat DNA preparation was 90%, and the median decrease in DNA load was 155 genome equivalents (geq) (P < 0.0001). For 327 specimens, PCR could be repeated on the primary DNA preparation. The sensitivity of PCR after storage was 95%, and the median decrease in DNA load was 13.5 geq (P < 0.0001). The specimens yielding negative results at repeat testing had a significantly lower median DNA load in the primary analysis than those with a repeat positive test (P < 0.0001). For 228 specimens, PCR could be performed both on the primary DNA preparation and after repeat DNA preparation. The median DNA load was lower after repeat DNA extraction than after repeat testing of the stored DNA extract (P < 0.0001). In conclusion, the M. genitalium DNA load as well as the detection rate decreased after storage. This was more pronounced in clinical specimens stored frozen than in stored DNA extracts, particularly in those with an initial low DNA load.