Raw Milk Illness Rate Is Low Per Serving Consumed and Per Consumer
Foodborne illness associated with raw milk is measured two ways in the epidemiological literature: per serving consumed, and per consumer. Both are calculated from the same underlying outbreak data. Both produce low absolute rates.
Per Serving Consumed
The per-serving figure comes from Costard et al. (2017), published in Emerging Infectious Diseases. The study examined dairy-related outbreaks in the United States from 2009 to 2014, covering four pathogens: Salmonella, Campylobacter, Shiga toxin-producing E. coli, and Listeria monocytogenes. It calculated an incidence rate ratio (IRR) of 840 per billion servings of unpasteurized dairy relative to pasteurized dairy, with a 95% credibility interval of 611 to 1,158.
Per-serving measurement is a standard epidemiological tool for comparing pathogen exposure across products consumed at very different volumes. Because raw milk is consumed in relatively small total volume by a small share of the population, while pasteurized milk is consumed by nearly the entire country, dividing illness counts by servings produces a large ratio even when the absolute illness counts are modest. The Costard authors are explicit about what the metric measures and what it does not.
Total dairy-related outbreak illness in the Costard dataset averages approximately 760 cases per year across both pasteurized and unpasteurized products combined.
Per Consumer
Dividing annual illnesses by the number of people who consume raw milk produces a different figure. The Costard study establishes that 3.2% of the U.S. population consumed raw milk as of 2014. Applied to the 2014 U.S. adult population of roughly 247 million, that yields approximately 7.9 million consumers and an illness rate of about 0.0096%, or 1 in 10,400. Other estimates using slightly different population bases produce a similar figure of around 1 in 14,000.
Per-consumer measurement reflects how often people who drink raw milk actually get sick. At 0.007% to 0.0096%, the per-consumer illness rate for raw milk is low in absolute terms. For context, per-consumer illness rates for leafy greensrun approximately 0.69%, chicken approximately 0.30%, and raw oysters approximately 0.23%.
Consumption Frequency Within the Consumer Population
The per-consumer figure above treats all past-year consumers as equivalent. Lando et al. (2022), published in the Journal of Food Protection and drawing on the FDA’s 2016 Food Safety Survey and 2019 Food Safety and Nutrition Survey, provides a frequency breakdown of that population.
The study found that 4.4% of U.S. adults, approximately 10.7 million people as measured across the 2016 and 2019 survey periods, reported drinking raw milk at least once in the prior twelve months. The Lando study’s Table 2 breaks that population down by reported frequency:
| Frequency | Share of raw milk drinkers | Estimated U.S. adults |
|---|---|---|
| Less than once per year | 23.1% | ~2,416,500 |
| A few times per year | 36.5% | ~3,959,800 |
| Subtotal: less than once per month | ~59.6% | ~6,376,000 |
| Once per month | 5.3% | ~574,000 |
| A few times per month | 8.1% | ~844,000 |
| Once per week | 4.9% | ~495,000 |
| Two to three times per week | 8.8% | ~990,000 |
| Daily | 9.5% | ~1,019,000 |
| Subtotal: once per month or more | ~36.6% | ~3,922,000 |
| Subtotal: once per week or more | ~23.2% | ~2,484,000 |
Approximately 60% of past-year raw milk consumers reported drinking it less than once a month. The standard per-consumer denominator includes that entire population. The Lando study notes this distinction matters for public health outreach, since infrequent and frequent consumers have different exposure profiles.
The 10.7 million figure is a 2016 to 2019 measurement. Raw milk sales have grown since then, and sales data from 2024 and 2025 suggest the current consumer base is larger. The frequency distribution would be expected to follow a similar pattern.
Illness Rate by Frequency Tier
Applying the Costard illness count of 761 per year against each frequency tier produces the following per-consumer rates:
| Consumer group | Estimated count | Annual illness rate | 1 in X |
|---|---|---|---|
| All past-year consumers (standard baseline) | 10,698,000 | 0.0071% | ~1 in 14,000 |
| Once per month or more | ~3,922,000 | 0.0194% | ~1 in 5,200 |
| Once per week or more | ~2,484,000 | 0.0306% | ~1 in 3,300 |
| USDA-ERS weekly figure (2014 to 2016) | 3,200,000 | 0.0238% | ~1 in 4,200 |
The USDA-ERS figure comes from Rhodes et al. (2019), which used the 2014 to 2016 American Time Use Survey and found that approximately 2% of at-home meal preparers, or 3.2 million people, consumed or served raw milk on a weekly basis. The FoodNet 2018 to 2019 population study found the same 2% figure among respondents in its catchment areas. Three independent datasets converge on roughly 3 to 3.2 million weekly consumers during the period covered by Costard’s illness estimates.
Limitations of Outbreak-Based Measures
Neither per-serving nor per-consumer rates derived from outbreak data are precise estimates of individual risk.
The illness numerator in both calculations is limited to outbreak-reported cases: illnesses that were formally investigated by public health departments and attributed to a specific food source. Sporadic illness, meaning cases that occur outside recognized outbreaks, accounts for the vast majority of all foodborne illness across every food category. Costard et al. note this limitation explicitly. The undercount applies to raw milk and to every comparison food.
The frequency-adjusted per-consumer figures carry an additional consideration. If infrequent consumers are unlikely to appear in the illness numerator, the illnesses are probably already concentrated among more frequent consumers. Restricting the denominator to that tier without a corresponding reduction in the numerator could overstate illness rates for regular consumers rather than refining the estimate.
What the data does establish across all denominators is a ceiling. Even under the most conservative scenario, applying all 761 annual outbreak-related illnesses against only the 2.5 million weekly consumers, the illness rate reaches 0.031%. Across every frequency tier examined, the per-consumer illness rate for raw milk remains well below comparable figures for leafy greens (approximately 0.69%), chicken (approximately 0.30%), and raw oysters (approximately 0.23%).
For a broader per-consumer comparison of raw milk illness rates against other commonly consumed foods, see Comparing Raw Milk Foodborne Illness Data with Other Foods. For a detailed breakdown of the 840x per-serving figure, see What “840 Times More Likely to Get Sick from Raw Milk” Actually Measures.
