Major errors (>0.5 ¿C) are produced in paleotemperature estimation with the C37 alkenone method when the amount of the diunsaturated or triunsaturated species, C37:2 and C37:3, respectively, approaches the limit of detection. These errors are more commonly encountered with C37:3 because of its higher adsorption to gas chromatographic (GC) columns. They are particularly relevant at low sedimentary C37 alkenone concentrations and may easily increase to temperature differences of 2¿--4 ¿C depending on the U37K' values and the ratio between the C37:3 signal and limit of detection. Mathematical expressions for the description of the analytical constraints derived from this detection limit problem are given. A quality assurance guideline based on sample dilution (three times) and calculation of the sea surface temperature error is proposed for assessment of reliable measurements. Reevaluation of literature data in the light of the equations reported in the present study suggest that some previously reported alkenone geochemical effects, such as variation in U37K' temperatures due to postdepositional oxidation, may reflect deviations in the GC measurements due to changes in C37 alkenone concentration. ¿ 2001 American Geophysical Union