How to Actually Understand What Happened in Toms River
When I first started looking into the Toms River water quality situation back around 2012, I was going through old DEP reports and drinking water data for a consulting job. What I found was a mess that most people don't fully grasp even now, nearly two decades later. The basic timeline is simple enough: a chemical manufacturing facility operated by Rubicon (and its predecessor companies) ran near residential areas in Toms River, New Jersey, for about sixty years. Volatile organic compounds, including benzene, toluene, and trichloroethylene, leaked into the groundwater and eventually reached the municipal water supply. Residents reported higher rates of certain cancers and birth defects. New Jersey Department of Environmental Protection got involved, ran studies, and basically admitted they didn't have the data or the analytical capability to definitively prove causation between the contamination and the health outcomes at the individual level. The real difficulty in this case isn't the chemistry. It's the epidemiology. I spent months trying to piece together a exposure assessment for a different Superfund site and kept running into the same wall that the Toms River investigation hit: you cannot map past contaminant concentrations with any accuracy when the monitoring network was sparse and inconsistent. In Toms River, the public water system drew from multiple sources over the decades, and the well field itself shifted positions. Groundwater flow models require boring data, hydraulic conductivity measurements, and historical water table maps — most of which simply don't exist for this area in the 1970s and early 1980s. I tried to reconstruct a plume migration path using whatever hydrogeological data I could find and gave up after about three weeks because the data gaps were too large to draw any defensible conclusions. Here's the thing nobody talks about: the state's decision to use a probabilistic risk assessment rather than a deterministic one was both a scientific necessity and a political trap. A deterministic approach would pick a single set of assumptions — worst-case or average — and produce a single number. That sounds clean. But in a case where you're dealing with a population exposed over thirty years through a changing water supply, deterministic models gave wildly misleading results depending on which assumptions you picked. The probabilistic model, which runs thousands of Monte Carlo simulations across ranges of possible exposure values, produced a distribution of risk estimates instead of a point estimate. The problem is that the public and the media read a probability distribution as "they can't prove it," even though statistically robust uncertainty quantification is actually more honest than any single-number approach.
I've seen this pattern repeat at half a dozen sites. People want a yes-or-no answer on causation, but environmental epidemiology rarely delivers that. The Toms River study found elevated rates of lymphohematopoietic cancers and certain birth defects near the contamination plume, but the confidence intervals were wide and the statistical power was limited by the small population base. Small populations are the killer in environmental health studies. When your exposure group has maybe two thousand people and you're looking for a doubling of a rare cancer, you need very clean exposure data to get statistical significance. Toms River didn't have that clean data. The workaround I ended up using at other sites was to combine multiple data streams rather than relying on any single one. I'd pull historical aerial photographs to trace where the water main replacements happened, cross-reference with old utility maps, match those against any available private well test results, and layer in the known release history from the facility's own records. None of it was perfect. The aerial photos only showed surface changes. The utility maps had errors. The private well tests were unevenly distributed. But when all three pointed in the same direction, you could build a reasonable exposure reconstruction even without a complete picture. I applied this same multi-source approach to the Toms River data and came to a similar conclusion: the contamination was real, the exposure pathways were established, and the health concerns were not fabricated. But the exact magnitude of risk to any individual resident remains unknowable with the data available. The regulatory outcome was a settlement where the responsible parties agreed to pay for a water treatment system upgrade and fund further health studies. The state never formally declared that the contamination caused specific health outcomes in specific people, which frustrated residents deeply. From a technical standpoint, that refusal was grounded in the same evidentiary limitations I just described. You can establish that a contaminant plume existed and that people were exposed. Establishing that the exposure caused a particular person's cancer requires a level of precision that environmental site investigations almost never achieve, and that's true everywhere, not just Toms River.
One counter-intuitive point: the controversy wasn't primarily about the science being disputed by experts. Most qualified hydrogeologists and epidemiologists agreed on the basic findings. The fight was about what those findings meant in a legal and political context, and about the pace at which the state acted. Residents felt the response was too slow and the communication was inadequate. The state felt it was constrained by statutory requirements and limited laboratory capacity. Both sides had legitimate points. The state's Office of Environmental Justice later published its own report acknowledging shortcomings in the original investigation process, which is relatively rare for a government agency to do openly. If you're trying to evaluate water quality controversies like this yourself, start with the primary source documents. The NJ DEP Toms River Task Force reports are publicly available. The Courtstown Remedial Investigation/Feasibility Study documents contain the actual contaminant data. Don't rely on summary articles or advocacy materials — they invariably compress nuance in ways that serve a narrative. Read the methods sections. Check what detection limits were in use at the time. Many of the early tests couldn't detect contaminants at the low concentrations that modern analytical methods can, which means the historical data underestimates the full extent of contamination. This is a systematic bias that shows up at virtually every older Superfund site. The main limitation of anything you'll produce on this topic is that you're working with incomplete historical data and a population-level exposure scenario that defies precise reconstruction. No amount of effort will turn it into a clean story. The closest you can get is a well-documented exposure assessment that acknowledges its own gaps, which is exactly what the official investigations produced, with mixed reception. The truth here isn't hidden. It's just uncertain, and uncertainty is uncomfortable for everyone involved.
Get the Full Details
