Free ionized calcium (fCa) is considered the gold standard for assessing calcium status in patients, but it is relatively expensive and is associated with several preanalytical and analytical error sources. We investigated the feasibility of using a reflex test that involves first measuring total calcium (tCa) and if out of reference range, then measure fCa, with expectation of reducing the number of fCa measurements. We used data from 1815 unique patients with concurrent measurement of fCa, tCa and albumin adjusted calcium (aCa). Patients were stratified by albumin level, and the association of fCa to tCa and aCa respectively was assessed with linear regression. The regression analysis showed the best linearity for tCa and aCa at albumin <35 g/L (R 2: 0.80–0.90), and the poorest at albumin >40 g/L (R 2: tCa 0.58; aCa 0.59). We examined the accuracy of hypo- and hypercalcemia classifications for tCa, aCa and the reflex test. aCa had more misclassifications of hypo- and hypercalcemia than tCa, with respectively 25% and 21%. Implementation of the reflex test would correct any false hypo- or hypercalcemia classified by tCa, leaving only false negative results corresponding to 9% of all tCa measurements. False negative results were on average 0.04 mmol/L above or below the reference range of fCa. Implementation of the reflex test reduces the number of fCa by 68% without major errors diagnosing hyper- or hypocalcemia.