<h3>Background</h3> Multiple myeloma (MM) is known to evolve from the premalignant precursor conditions Monoclonal Gammopathy of Undetermined Significance (MGUS) and Smoldering Multiple Myeloma (SMM). The exact mechanisms driving this progression are still not completely understood. The continuum between MGUS-SMM-MM provides the opportunity to investigate its evolution. However, collecting paired/serial samples from low and intermediate risk precursors progressing to MM are significantly hindered by the (i) only incidental diagnosis of asymptomatic precursors, (ii) invasive bone marrow (BM) sampling, (iii) very low number of aberrant BM plasma cells (PCs), and (iv) rather irregular disease follow-up. <h3>Methods</h3> In this study, a unique retrospective collection of paired BM samples was available because of an effective biobanking effort. We had access to 68 archival diagnostic May-Grünwald–Giemsa (MGG)-stained BM smears from 21 progressing low to intermediate risk myeloma precursor patients, 19 MGUS and 2 SMM, with a median time to progression of 6 years. DNA was extracted from these BM smears and Next Generation Sequencing (NGS) was performed using a custom targeted capture-based sequencing panel (Illumina) including coding exons or hotspots of 81 selected myeloma-related genes with the aim to study the evolution of single nucleotide variants (SNVs) and short insertions and deletions (indels). The pooled libraries were paired-end sequenced on a MiSeq instrument (Illumina). <h3>Results</h3> Data was analyzed with Local Run Manager and annotated with VariantStudio. After filtering, only exonic nonsynonymous and loss-of-function variants with a coverage >30 and a variant allele frequency (AF) >1% were retrieved. A variant detected in the MM phase but not in the prestage of that patient was manually inspected in the data visualization tool of Integrated Genome Viewer (IGV) to assess its presence. With a mean coverage of 636x of the targeted captured region the sequencing depth was sufficiently high to detect low burden variants down to an AF of 1%. A median of 2 variants per patient (range 0 to 5 with a total of 38 variants) was detected. While in 19 patients at least one variant was detected in the MM phase, in 2 patients not a single variant was detected in any of the 81 genes. Interestingly, the majority of variants in MM could already be detected at low AFs in the BM smears sampled in the precursor phase, even many years before progression. The reason why some variants were not detected in a prestage is likely due to sensitivity issues (BM PCs ≤2.5%). The median time of detection of a variant in the precursor stage BM was 49 months (>4 years; range 10 to 105 months) prior to MM progression. <h3>Conclusion</h3> In conclusion, targeted sequencing of unsorted BM smears from myeloma precursor conditions can already provide relevant insights into the behavior of mutant clones. Paired sample analysis can reveal the genetic architecture of somatic variation in the prestage even with low proliferative PCs in the BM.