Archaeomagnetic field intensity in Finland (Lat. 61°N, Long. 22° E) has been determined for the past 6400 years with the Thellier technique of bricks, potsherds and baked clays. The normalized archaeointensity curve for Finland, based on 23 mean values, shows an increase from ∼4360 BC to the maximum at AD 500, after which it decreases to the present value. The high intensity at AD 500 as found in Finland, but not in Bulgaria and Japan, is not a consequence of the applied Thellier technique since we are able to reproduce the known field intensities in the laboratory, and some of the dated bricks yield values which are in broad agreement with Nurmijarvi Magnetic Observatory data. We also show that variations in grain size of the magnetic carriers, cooling rate or magnetic refraction are unlikely to cause systematic errors in intensity larger than a few percent. Based on results of self-made bricks we show that the archaeointensity, however, depends on the angle between the laboratory and ancient field directions. If these are not aligned errors up to ±10% in archaeointensity can occur and are probably caused by anisotropy of remanence or by other factors not yet understood. The high intensity at AD 500 in Finland can be modelled by a radial nondipole field source at the core-mantle boundary producing enhanced field at higher latitudes (Finland) and relatively weaker fields at lower latitudes (e.g. Bulgaria and Japan). Extrapolation of the present field (IGRF 1990) back in time shows similar field behaviour.