Many sunspot number series exist suggesting different levels of solar activity during the past centuries. Their reliability can be assessed only by comparing them with alternative indirect proxies. We test different sunspot number series against the updated record of cosmogenic radionuclide 44 Ti measured in meteorites. Two bounding scenarios of solar activity changes have been considered: the HH-scenario (based on the series by Svalgaard and Schatten), in particular, predicting moderate activity during the Maunder minimum, and the LL-scenario (based on the R G series by Lockwood et al.) predicting moderate activity for the 18th-19th centuries and the very low activity level for the Maunder minimum. For each scenario, the magnetic open solar flux, the heliospheric modulation potential and the expected production of 44 Ti were computed. The calculated production rates were compared with the corresponding measurements of 44 Ti activity in stony meteorites fallen since 1766. The analysis reveals that the LL-scenario is fully consistent with the measured 44 Ti data, in particular, recovering the observed secular trend between the 17th century and the Modern grand maximum. On the contrary, the HH-scenario appears significantly inconsistent with the data, mostly due to the moderate level of activity during the Maunder minimum. It is concluded that the HHscenario sunspot number reconstruction significantly overestimates solar activity prior to the mid-18th century, especially during the Maunder minimum. The exact level of solar activity after 1750 cannot be distinguished with this method, since both H-and L-scenarios appear statistically consistent with the data.