EN
We deal with the test of the general relativistic gravitomagnetic Lense-Thirring effect currently being conducted in the Earth’s gravitational field with the combined nodes Ω of the laser-ranged geodetic satellites LAGEOS and LAGEOS II. One of the most important sources of systematic uncertainty on the orbits of the LAGEOS satellites, with respect to the Lense-Thirring signature, is the bias due to the even zonal harmonic coefficients J
ℓ of the multipolar expansion of the Earth’s geopotential which account for the departures from sphericity of the terrestrial gravitational potential induced by the centrifugal effects of its diurnal rotation. The issue addressed here is: are the so far published evaluations of such a systematic error reliable and realistic? The answer is negative. Indeed, if the difference ΔJ
ℓ among the even zonals estimated in different global solutions (EIGEN-GRACE02S, EIGEN-CG03C, GGM02S, GGM03S, ITG-Grace02, ITG-Grace03s, JEM01-RL03B, EGM2008, AIUB-GRACE01S) is assumed for the uncertainties δJ
ℓ instead of using their more-or-less calibrated covariances $$
\sigma _{J_\ell }
$$, it turns out that the systematic error δμ in the Lense-Thirring measurement is about 3 to 4 times larger than in the evaluations so far published based on the use of the covariances of one model at a time separately, amounting up to 37% for the pair EIGEN-GRACE02S/ITG-Grace03s. The comparison among the other recent GRACE-based models yields bias as large as about 25–30%. The major discrepancies still occur for J
4; J
6 and J
8, which are just to which the zonals the combined LAGEOS/LAGOES II nodes are most sensitive.