Cell division happens during the M phase (mitosis or meiosis), the final stage of the cell cycle. Interphase—G1, S, and G2—occupies roughly 90–95% of a typical cycling cell's time and is when the cell grows and replicates its DNA. The M phase is short by comparison: mitosis in a typical mammalian cultured cell runs about 1 hour, with cytokinesis completing physical separation.

The question sounds simple, but the honest answer has layers a thin explainer misses. Division is not one event but a tightly gated sequence with three irreversible transitions. It is also not purely an M-phase story: decisions made in G1 and G2 determine whether division happens at all. What follows separates the chromosome segregation events of mitosis from the cytoplasmic splitting of cytokinesis, places both on a measured timeline, and covers the edge cases—endoreplication, binucleation, asymmetric division—that complicate the textbook picture.

Molecular machinery of the M phase

The M phase is best understood as an ordered hand-off of control between two oscillator systems. The first is the cyclin B–CDK1 complex, which accumulates through G2 and triggers mitotic entry. The second is the anaphase-promoting complex/cyclosome (APC/C), a ubiquitin ligase that dismantles CDK1 activity and locks the cell into exit. Confusing the two produces most of the errors in popular accounts.

CDK1 activation requires removal of inhibitory phosphates by the phosphatase CDC25. This creates a positive-feedback loop: active CDK1 activates more CDC25 and inhibits the kinase WEE1 that installs the inhibitory phosphates. The loop is switch-like, which is why mitotic entry is abrupt rather than gradual. A cell does not drift into mitosis; it flips.

Chromosome segregation depends on a separate quality-control device. The spindle assembly checkpoint (SAC) senses unattached kinetochores—protein structures on each chromatid that bind spindle microtubules. While even one kinetochore remains unattached, the SAC inhibits the APC/C. Once every kinetochore is attached and under tension, the SAC silences, the APC/C activates, and separase cleaves cohesin, the ring complex holding sister chromatids together. Chromosomes move to opposite poles.

Cytokinesis is mechanically distinct. In animal cells, an actomyosin contractile ring assembles at the equatorial cortex, constricts, and is severed by the ESCRT-III machinery. The abscission step is the true final cut, and it can fail—producing the binucleate and multinucleate cells discussed below.

In plain terms: mitosis moves the cargo, and cytokinesis cuts the container. Neither step is automatic; both are licensed by prior events.

How long each phase actually takes

Durations vary enormously by cell type and species, so any single number is a model, not a rule. The table below gives representative values for a proliferating mammalian cell line in culture (such as HeLa) and, where known, for human tissues in vivo. Treat these as order-of-magnitude guides.

Phase Main event Typical duration (cultured mammalian cell) Notes on in vivo variation
G1 Growth; restriction point decision 10–12 hours Highly variable; can last years in quiescent cells
S DNA replication 6–8 hours Firing of replication origins is staggered
G2 Repair and mitotic preparation 3–4 hours Checkpoint-dependent
Prophase/prometaphase Condensation; nuclear envelope breakdown 20–40 minutes Envelope breakdown marks the mitosis/cytokinesis boundary
Metaphase Chromosome alignment; SAC satisfaction 10–20 minutes Prolonged by even one unattached kinetochore
Anaphase Sister chromatid separation 5–10 minutes Fast and largely irreversible
Telophase plus cytokinesis Nuclear reassembly; ring constriction and abscission 20–60 minutes Abscission is the final committed step

Two caveats matter. First, M phase occupies roughly 5–10% of a cycling cell's total cycle time—consistent with the 90–95% interphase figure above—but that fraction shifts with cell type. Second, embryonic cleavage divisions can compress the entire cycle to under 30 minutes by trimming G1 and G2 almost entirely. The M phase itself does not shorten proportionally, which is why early embryos spend a larger fraction of each cycle in division.

Edge cases: when division departs from the textbook

Endoreplication. Some cells replicate DNA through S phase but skip M phase entirely, producing polyploid nuclei. Megakaryocytes, which generate platelets, reach DNA contents of 16N to 64N through repeated endomitotic cycles. Liver hepatocytes and placental trophoblast giant cells do the same at lower ploidy. These cells divide their genome without dividing themselves.

Binucleation and failed abscission. If the contractile ring or ESCRT-III machinery fails after karyokinesis, the cell completes mitosis but not cytokinesis. The result is a binucleate cell. Hepatocytes in adult liver commonly show binucleation, and its frequency changes with age and injury. This is a normal physiological outcome in some tissues, not always a defect.

Asymmetric division. Stem cells can orient the mitotic spindle and the contractile ring so that daughter cells receive unequal amounts of fate determinants. This is division, but not symmetric division—a distinction that matters whenever "cell division" is treated as synonymous with "doubling."

Meiosis. Germ cells divide twice with only one S phase. Meiosis I separates homologous chromosomes; meiosis II separates sister chromatids. The M-phase label applies to both divisions, but their regulation differs—cohesin is released in a stepwise fashion, and the SAC behaves differently across the two meiotic divisions.

A counterintuitive professional caveat: the SAC is not a perfect fidelity device. It can be satisfied by inappropriate kinetochore attachments under some conditions, and SAC weakening contributes to chromosomal instability in tumors. Checkpoint satisfaction is not the same as accuracy.

Common misconceptions, corrected

The most frequent error is equating "cell division" with mitosis alone. Mitosis is chromosome segregation; cytokinesis is physical separation; the two can and do dissociate. A cell that completes mitosis without cytokinesis has divided its nucleus but not itself.

The second error is treating interphase as "resting." Interphase contains the S phase, where the entire genome is duplicated, and the G1 and G2 checkpoints, where division is licensed or blocked. Roughly 90–95% of cycle time is spent here, and most regulation of whether division occurs happens in interphase.

The third error is assuming the G1/S restriction point and the G2/M checkpoint are interchangeable gates. They are not:

G1/S checkpoint: assesses cell size, nutrient status, and DNA damage before replication commitment. Intra-S checkpoint: slows or stalls replication forks in response to damage. G2/M checkpoint: prevents mitotic entry while damage remains unrepaired. Spindle assembly checkpoint: blocks anaphase until kinetochores are attached and tensioned. Cytokinesis checkpoint (abscission delay): holds the final cut when chromatin is trapped in the intercellular bridge.

Red flags that indicate a question or claim is over-simplified:

Any statement that "the cell cycle is G1, S, G2, M" without noting that M comprises two mechanistically separate events. Duration figures presented without cell type or culture condition. Claims that a specific drug "stops cell division" without specifying which transition it blocks. Use of "resting phase" for G0 as if quiescent cells are inert; they are metabolically active. Percentages of cells in each phase quoted from a single study as universal values.

Practical actions: measuring and interpreting phase timing

These steps help anyone—student, lab worker, or clinician reading a pathology report—interpret cell-cycle data correctly.

Specify the measurement method before quoting durations. Flow cytometry after DNA staining gives population-level G1/S/G2-M fractions, not individual cell times. Pulse-chase labeling with thymidine analogs (BrdU or EdU) measures S-phase transit. Live imaging gives direct per-cell timing. Choose the method that answers the question asked.

Convert percentages to times only with a measured total cycle time. A population that is 60% G1 and 10% G2/M does not tell you how long M phase lasts unless you know the doubling time.

Check whether the cell type is endoreplicating or binucleate before applying diploid assumptions. DNA content measured by flow cytometry will show 4N and 8N populations that are not G2/M cells.

For mitotic index calculations, define the marker. Phospho-histone H3 (Ser10) marks mitotic chromatin; it does not mark cytokinesis. Pair it with a cytokinesis marker if separation is the outcome of interest.

When interpreting a pathology report that states a mitotic count, note that counts reflect cells fixed in a narrow window—typically prophase through telophase—and underestimate total division frequency. Counts are also method- and fixation-dependent.

If timing matters for an experiment, pilot the checkpoint response. Damaging agents shift cells out of M phase on a schedule that depends on dose and checkpoint proficiency, not on a fixed clock.

Limits of what current evidence supports

Several widely repeated numbers should be treated cautiously. The "90–95% interphase" figure is a reasonable summary for rapidly proliferating cultured cells, but it was not established as a universal constant and does not describe quiescent or slowly cycling tissues. Similarly, the phase-by-phase durations in the table above come from a small number of well-studied cell lines. Extrapolating them to human tissues in vivo requires independent measurement, and such measurements are sparse for most organs.

The deeper limitation is conceptual. "During which stage does division happen" assumes division is a discrete event at a definable moment. In reality, division is a process with at least three committed transitions: SAC satisfaction, separase activation, and abscission. Each can be arrested independently, and the cell can persist in an intermediate state—post-mitotic but pre-cytokinetic—for extended periods. Any answer that names a single stage is a useful approximation, not a complete mechanistic account.

This article is general educational information about cell biology. It is not a diagnosis, treatment plan, or clinical recommendation, and it does not describe patient-specific testing or therapy. Anyone interpreting cell-cycle data in a clinical or research context should consult a qualified local pathologist, oncologist, or laboratory director.