Treatment‐free remission (TFR) by tyrosine kinase inhibitors (TKI) discontinuation in patients with deep molecular response (DMR) is a paramount goal in the current chronic myeloid leukemia (CML) therapeutic strategy. The best DMR level by real‐time quantitative PCR (RT‐qPCR) for TKI discontinuation is still a matter of debate. To compare the accuracy of digital PCR (dPCR) and RT‐qPCR for
BCR‐ABL1
transcript levels detection, 142 CML patients were monitored for a median time of 24 months. Digital PCR detected
BCR‐ABL1
transcripts in the RT‐qPCR undetectable cases. The dPCR analysis of the samples, grouped by the MR classes, revealed a significant difference between MR
4.0
and MR
4.5
(
P
= 0.0104) or MR
5.0
(
P
= 0.0032). The clinical and hematological characteristics of the patients grouped according to DMR classes (MR
4.0
vs MR
4.5‐5.0
) were superimposable. Conversely, patients with dPCR values <0.468
BCR‐ABL1
copies/µL (as we previously described) showed a longer DMR duration (
P
= 0.0220) and mainly belonged to MR
4.5‐5.0
(
P
= 0.0442) classes compared to patients with higher dPCR values. Among the 142 patients, 111 (78%) discontinued the TKI treatment; among the 111 patients, 24 (22%) lost the MR
3.0
or MR
4.0
. RT‐qPCR was not able to discriminate patients with higher risk of MR loss after discontinuation (
P
= 0.8100). On the contrary, according to dPCR, 12/25 (48%) patients with
BCR‐ABL1
values ≥0.468 and 12/86 (14%) patients with
BCR‐ABL1
values <0.468 lost DMR in this cohort, respectively (
P
= 0.0003). Treatment‐free remission of patients who discontinued TKI with a dPCR <0.468 was significantly higher compared to patients with dPCR ≥ 0.468 (TFR at 2 years 83% vs 52%
P
= 0.0017, respectively). In conclusion, dPCR resulted in an improved recognition of stable DMR and of candidates to TKI discontinuation.