We propose a new paleointensity method based on normalization by isothermal remanent magnetization derivative vs. alternating field (REM′ method). It provides an estimate of the absolute paleointensity with an uncertainty of about a factor two. Contrary to methods using normalization by total isothermal remanent magnetization or anhysteretic magnetization, it is applicable to multicomponent magnetizations. Artificial or natural isothermal remanent magnetizations can be recognized, and relaxation of natural remanent magnetization can be taken into account. It is applicable for magnetizations carried by magnetite, FeNi alloys and pyrrhotite. The REM′ method is of particular interest to estimate paleointensities in meteorites, as these materials are often characterized by complex multicomponent magnetization of various origins (impact magnetization, thermoremanence, isothermal remanent magnetization) and metastable magnetic minerals not suitable for Thellier experiments. However, like other paleointensity methods, the REM′ method underestimates the true paleointensity for meteorite samples with magnetizations that are heterogeneous on spatial scales below the sampling scale. For L ordinary chondrites, an upper limit of 1 μT is proposed for the paleofield. Tentative paleofield estimates in the 0.05–0.5 μT range are proposed for LL ordinary chondrites, which is much lower than previous results. The first paleofield estimates for Rumuruti chondrites indicate the presence of magnetic fields around 6 μT during the last major impact on their parent body. Aubrites and HED achondrites paleofields of at least 10 μT may suggest the existence of a dynamo field during the cooling of their parent body. Results from Martian meteorites are scattered between 1 and 24 μT and may represent the crustal magnetic field of the planet after dynamo shutdown. Carbonaceous chondrites provide contrasting results, with possible evidence of strong field (mT) processes that clearly require additional studies.