Background: Traditionally, biomarker ascertainment occurred once for patients (pts) with mCRC. An advantage of ctDNA is the ease of repeated assessments. However, real-world evidence about the value of serial ctDNA in revealing actionable alterations is needed. Methods: We retrospectively evaluated 3350 pts with mCRC and a Guardant360 ctDNA assay (Guardant Health) revealing ≥1 somatic alteration who underwent ≥1 subsequent Guardant360 test to compare detection of mutations (mut: single nucleotide variants (SNVs) and indels), amplifications (amps), fusions, microsatellite instability (MSI), and blood tumor mutational burden (bTMB). Variants were filtered to pt-specific levels of detection by limiting variants across assays to those above a 0.1% mutant allele frequency (MAF) threshold on the serial test with the lowest somatic MAF for that pt. Clonal muts were defined as those at ≥10% of max MAF per sample. Results: A total of 9130 assays (mean of 2.7 assays/pt) occurring a median of 165 days apart were evaluated. Among 1476 pts initially with no alteration in MAPK pathway genes (RAS and EGFR SNV or indel, BRAF V600E SNV, or ERBB2/MET amplifications), 382 (25.8%) acquired a MAPK alteration on their second test. More gains were clonal than subclonal (231:151, 60.5%:39.5%). In pts with a subsequent assay after an acquisition (N = 80), 12 (15.0%) clonal and 15 (18.8%) subclonal acquisitions disappeared in the third assay, a median of 150 days later. Among alterations with a therapy, KRAS G12C/D, BRAF V600E, and ERBB2 amps acquisitions occurred at any later assay in 84/1476 (5.7%), 29/1476 (2%), and 21/1476 (1.4%) pts without an initial MAPK alteration, respectively. Of these, 84/134 (62.6%) emerged without another concurrent MAPK alteration, 61/84 (72.6%) of which are subclonal. Of 92 fusions noted in 86 pts, 87/92 (94.5%) were subclonal and only 28/92 (30.4%) were initially present. The majority of fusions were acquired de-novo subclonal fusions (58/64, 90.6%) but 2 pre-existing subclonal fusions subsequently became clonal. Among pts evaluable for MSI, 56/3030 (1.8%) were initially MSI-H and 30 (1%) subsequently had MSI detected on a future assay. New MSI detection was more common in pts with DNA repair muts (BRCA1/BRCA2/ATM/CHEK2/MLH1/RAD51D) on an initial assay (OR 7.52, 95% CI 3.39-16.69, P < 0.0001). Among pts evaluable for bTMB, 60/1387 (4.3%) initially had bTMB≥20 muts/Mb, and 256/1327 (19.3%) subsequently rose above 20 muts/Mb on a future assay, a median of 417 days after initial assay. Rising bTMB associated with rising max somatic MAF (Spearman rho = 0.50, P < 0.0001). Pts with DNA repair muts on an initial assay were more likely to have bTMB rise to ≥20 muts/Mb (OR 2.58, 95% CI 1.81-3.68, P < 0.001). Conclusions: In this large mCRC cohort, serial ctDNA appears to be a feasible approach to identify acquired alterations with therapeutic implications.