How Long Does It Take a Dead Body to Decompose? The Cases the Usual Answer Does Not Fit

On the ground in a temperate summer, a human body usually reaches advanced decay within one to three weeks and skeletonization within a few months; buried at a meter, it can take years. Forensic practice counts accumulated degree-days rather than days, meaning days multiplied by mean daily temperature in degrees Celsius. The standard model, published by Megyesi, Nawrocki and Haskell in the Journal of Forensic Sciences in 2005, reads degree-days off appearance as ADD = 10^(0.002 × TBS² + 1.81), where TBS is a decomposition score running from 3 to 35. It explains about 84% of the variation in its own data and carries a prediction interval of ±388.16 degree-days, which at a steady 20 °C is a little over nineteen days each way.

The usual answer is a ratio from 1858

The figure most people meet descends from Johann Ludwig Casper, a Berlin physician who around 1858 proposed a fixed ratio across three settings. One week of putrefaction in open air equals two weeks in water equals eight weeks buried in soil, at equal temperature. Teaching texts still repeat it, with the thermal envelope added later: putrefaction runs fastest between about 21 °C and 43 °C, stalls near freezing, stops above 48 °C.

Stage tables sit on top of that ratio. Galloway and colleagues reviewed 189 southern Arizona cases for the Journal of Forensic Sciences in 1989 and put bloating at two to seven days after death. Summaries of the same dataset give summer skeletonization inside six weeks against roughly four months in winter, where summer air passes 38 °C. A factor of three between seasons, same desert.

What the clock actually measures

Sixty days at 10 °C and twenty days at 30 °C are both 600 accumulated degree-days, and to a first approximation the same amount of decomposition. This is why how long has no answer without how warm.

Megyesi's team turned that into a working equation. An examiner scores the head, the trunk and the limbs separately, sums them into a Total Body Score, and reads accumulated degree-days off the curve. The fit looks strong, Total Body Score accounting for roughly 84% of the variance across their sample.

I used to let a number like that travel alone. Until around 2017 our butter series went out as a headline per-capita figure, the interval buried in an annex. A member organization quoted that headline in a submission; the next year's figure moved by more than the change they claimed. The interval now shares a sentence with the estimate, or the estimate stays in.

Where those 68 cases came from

The 84% is a fit to 68 cases. Later scrutiny of the same dataset found that in 71% of them the date of death was not documented at all; it had been estimated from insect evidence. The regression was fitted against another estimate rather than calendar dates.

Moffatt, Simmons and Lynch-Aird set out the arithmetic in 2016 in the same journal, identifying errors in rounding, in the temperature scale and in the regression itself, and noting that the original equation could return negative degree-day values. Refitting on the more reliable portion produced r² = 0.91 with markedly narrower confidence intervals; a 2023 commentary recomputed on the fifteen data points that group had used. Two symptoms survive in the published version: it returns 49 degree-days for a Total Body Score of zero, which describes a fresh body, and the data contains no case below 55.9 degree-days.

I recognize the shape of that because I produced one. In 2012 I reported a fall in household sugar that was really a change in the diary form, where the coding list had dropped the prompt for sugar in hot drinks. Fourteen months passed before I found it. It cost a reissued table and an uncomfortable meeting. Before reading a series now, I find out how each value got into it.

What happened when the model met known dates

A validation study placed 28 donated bodies at three American human decomposition facilities, in sun and shade, four times a year, with dates of death and degree-days genuinely known. Nine experienced observers scored them, and the scoring held up: intraclass correlation coefficients of 0.975 for the trunk, 0.959 for the head and 0.940 for the limbs, all at p < 0.001.

Agreement with the calendar was another matter. Compared at 100, 300, 500 and 1,000 actual degree-days, the mean estimates did not correlate well with the true values at any of the four levels, and the error was largely unpredictable except that it grew with time since death. Strong agreement between scorers with weak agreement with reality is familiar to anyone who runs a survey. The instrument is consistent, and consistency says nothing about what it measures.

This is the edge of what I can offer. I have never scored a body, never attended a recovery, and cannot tell you how a limb score behaves at dusk in long grass. What I can vouch for is what happens to a regression when most of its dependent values are themselves estimates.

The temperature record is a second estimate stacked on the first

Every degree-day figure rests on a temperature history nobody recorded at the body. The accepted repair is to log temperature where the remains lay and regress those readings against the nearest weather station, correcting its archive backwards. Lutz and Amendt's applied primer on temperature in forensic entomological casework gives the limits: a station within 15 km, scene measurement across 3 to 10 consecutive days, and a mean station-to-scene difference under 5 °C.

Done properly, the order matters:

1. Log at the scene, in the position the remains occupied, for at least three consecutive days.
2. Regress that series against the same period from the reference station.
3. Apply that relationship backwards across the archive, reporting the residual spread.

Mostly it is not done. The same primer's survey of published casework found retrospective correction applied in roughly 13% of indoor cases and about 6% of outdoor ones. Correction is also not free. Dourel and colleagues, writing in Psyche in 2010, simulated it across seven meteorological stations, three exposure conditions and three seasons using 5, 10 and 15-day correlation windows, and found no consistent benefit over raw station data; Archer's earlier Australian work found corrected series described the site better while the decomposition estimate sometimes worsened. The iButton loggers commonly used are specified to ±0.5 °C, and an offset that size held across a long interval moves a minimum postmortem interval by about a day.

Burial changes the question, and three things change it further

Depth, and the five hours before it

Bachmann and Simmons buried 60 rabbit carcasses at 35 cm in 2010, in two halves. Thirty were left accessible to insects for five hours before burial; thirty were held in sealed bags at 4 °C and went into the ground at the same moment. Exhumed at 50-degree-day intervals, the exposed group decomposed roughly 30% faster, at p < 0.001. Five hours above ground, then months of identical conditions, and the curves never converge.

Depth governs whether that window stays open. Rodriguez and Bass buried six unembalmed cadavers in unlined trenches of varying depth at Knoxville, exhuming them between one month and one year later and recording air, soil and cadaver temperatures daily. They watched female flies lay eggs on the soil surface after heavy rain, the larvae working down toward the remains through a barrier that rain had opened.

Soil moisture is not a dial

Carter, Yellowlees and Tibbett buried rat cadavers of about 18 g in 500 g portions of three contrasting Queensland soils, calibrated to matric potentials of −0.01, −0.05 and −0.3 MPa and incubated at 22 °C. In sand and loamy sand, wetter meant faster. In medium clay it did not; the wettest treatment passed the optimum and slowed decomposition. Moisture modifies the relationship between temperature and decay rather than adding to it. A degree-day figure quoted without a soil description does less work than it appears to.

What a coffin actually does

Forbes, Stuart and Dent ran twelve-month burial microcosms in Forensic Science International in 2005, using pig adipose tissue in damp loamy sand at about 22 °C, three replicates each. Tissue buried directly in soil converted entirely to adipocere at 94% saturated fatty acids. In mock chipboard coffins it produced only scanty adipocere at 79–84% saturated, under a fungal mat of four species. The wood had not disintegrated after twelve months, but the coffins were no longer sealed.

Mant's post-war exhumations had already reported bodies in coffins breaking down faster than bodies placed straight into soil, the slightly aerobic interior favoring ordinary decay. The clothing result is sharper. Cotton disintegrated completely inside twelve months, matching Morse and Dailey's estimate of under ten months for cotton in soil, while polyester survived intact and left tissue at over 97% saturated fatty acids.

An explainer and an estimate are different objects

| | Reads time from | Reports | Fails when | |---|---|---|---| | Stage-range explainer | Calendar days in a named setting | One range, such as "weeks to months" | The actual setting differs from the one the table was built on | | Total Body Score and degree-day model | Degree-days inferred from appearance | A point estimate with a wide interval | Scene temperature history is unknown, or the body falls outside the fitted range | | Case-specific estimate | Scene logger, insect evidence, scene context | A bounded interval with stated assumptions | Access, temperature or disturbance went undocumented |

The strongest argument against everything above is practical. An investigator standing over remains needs a number this week, and an interval spanning a season is not a number; a point estimate that is roughly right and available on Tuesday beats an honest range nobody can act on. That is true, and it is why the stage tables persist. The framing is what I would not concede. An interval is not a refusal to answer, it is the part of the answer that survives cross-examination.

Questions people actually ask

How long does a body take to decompose to a skeleton?

On the surface in warm weather, weeks. Galloway's Arizona cases reached skeletonization inside six weeks in summer and nearer four months in winter. Buried at a meter, the same process takes years. Forensic estimates express this as accumulated degree-days rather than calendar time, because temperature rather than elapsed days sets the rate.

How long does a body take to decompose after burial?

Casper's 1858 ratio still frames it: one week of putrefaction in air equals about eight weeks buried, at the same temperature. Depth and insect access matter more than the calendar. Rabbit carcasses given five hours of insect exposure before burial at 35 cm decomposed roughly 30% faster than sealed controls.

How long does a body take to decompose in a coffin?

Faster than most people assume. Mant's post-war exhumations and a 2005 Forensic Science International burial experiment both found soft tissue breaking down more readily in coffins than in direct soil burial, because the slightly aerobic interior discourages adipocere. Mock coffins in that study were no longer sealed after twelve months.

How long does decomposition take in hot weather?

Faster, up to a limit. Putrefaction runs fastest between roughly 21 °C and 43 °C and effectively halts above 48 °C. In southern Arizona, where summer air passes 38 °C, bloating appears two to seven days after death, though extreme aridity can arrest the process into mummification instead.

What happens 30 minutes after death?

Pallor mortis is complete, the skin having paled within 15 to 25 minutes as capillary circulation collapses. Lividity may just be visible in dependent areas; it can appear from about 20 minutes and fixes at 8 to 12 hours. Cooling has begun, near 2 °C in the first hour. Rigor starts later, around two hours.

Which local temperature and access conditions were documented by the investigator?

Ask for four things: whether a logger was placed where the remains lay, how many days it ran, how far the reference weather station sits, and whether insects had access throughout. Accepted practice is a station within 15 km, 3 to 10 days of scene logging, and a mean difference under 5 °C.