This study presents the first archeointensity results from Northeast Brazil obtained from 14 groups of architectural brick fragments sampled in the city of Salvador, Bahia State (13°S, 38.5°W) and dated between the middle of the XVIth century and the beginning of the XIXth century. The dating is ascertained by historical documents complemented by archeological constraints, yielding in all cases age uncertainties of less than 50 years. Analyses were carried out using two experimental protocols: 1 — the "zero field-in field" version of the classical Thellier and Thellier method as proposed by Coe (TT-ZI), including partial thermoremanent magnetization (pTRM) and pTRM-tail checks, and 2 — the Triaxe procedure involving continuous high temperature magnetization measurements. Both TRM anisotropy and cooling rate effects were taken into account for the intensity determinations. The cooling rate effect was further explored for the TT-ZI protocol using three increasing slow cooling times (5 h, 10 h and 25 h) between 450 °C and room temperature. Following archeological constraints, the slowest cooling time was retained in our study, yielding decreases of the raw intensity values by 4% to 14%. For each fragment, a mean intensity was computed and retained only when the data obtained from all specimens (between 2 and 6) satisfied a coherence test at ∼ 5%. A total of 57 fragments (183 specimens) was considered for the computations of site-mean intensity values, with derived standard deviations of less than 8% of the corresponding means. When separately computed using the two experimental techniques, the site-mean intensity values always agree to within 5%. A good consistency is observed between intensity values of similar or close ages, which strengthen their reliability. Our data principally show a significant and continuous decrease in geomagnetic field intensity in Northeast Brazil between the first half of the XVIIth century and the XXth century. One result dated to the second half of the XVIth century further suggests that the geomagnetic field intensity reached a maximum around 1600 AD. This evolution is in good agreement with that expected in the city of Salvador from the available global geomagnetic field models. However, the accuracy of these models appears less well constrained between ∼ 1550 AD and ∼ 1650 AD.