Six years ago I arrived at the University of California at San Francisco to learn how to experiment on mice. My Ph.D. research in immunology had examined one part of the development of white blood cells in mammals, but on dishes and in test tubes. I wanted to learn how to play games with developing cells in an intact mouse, the experimental system most immunologists preferred.

Within a few months, I asked my new supervisor to switch back to what was more familiar. At the time, I justified the change to myself by reasoning that I had to. I wasn’t good enough at the technique I was supposed to be learning how to use, and my boss wanted results, not fumbling. Plus, I discovered that our laboratory would be moving to Germany, and cumbersome mouse work would take a long time to resume after the move. But another reason drove me away from working with live mice. I had to kill them.

My original project was to take a growing thymus out of an embryonic mouse and infect it with a genetically engineered virus. T cells, white blood cells that control the adaptive immune system, develop in the thymus. The viruses carried genes that I hoped would nudge T cell development in different directions. The laboratory’s mouse technician, a cheerful blonde Californian, set up mouse matings at my request. I’d monitor the pregnant mothers and then “sacrifice” them at the right time and dissect their embryos.

Some of my co-workers preferred to suffocate mice with carbon dioxide. A few minutes in a bucket with a chunk of dry ice did the trick. Juan, a Spanish specialist in anatomy who had been in the lab many years, showed me a way that was faster than that indirect, peaceful method. A quick death by cervical dislocation is better, he said. Hold the mouse by its tail, he demonstrated, let it grab the bars of its cage on the outside, and pull on its tail while pushing on the back of its neck with a spatula. Juan could do everything in one smooth motion. Hesitation made the task more difficult because the mice squirmed furiously.

Imitating Juan, I soon became accustomed to killing mice but dissecting the embryos was another obstacle course. A mouse embryo fourteen days after conception, the right age to infect the thymus, is about as large as a thumbnail. The thymus itself, a soft pair of 2 mm wide balls, lies in the chest between the throat and the heart. Little distinguishes it from the surrounding tissue. If I poked the heart with my forceps, oozing blood would obscure my prize. Frequently I’d mangle the embryo beyond recognition or destroy what I was trying to extract. Sometimes miscommunication with the mouse technician left me with embryos that were the wrong age, too small or too large. What a waste, I thought.

Our research on mice seemed innocuous compared with the images animal rights protesters use to promote their cause. While I was learning these techniques, a group of activists protested outside on Parnassus Avenue. They didn’t care about my work specifically. They focused on researchers at UCSF who deafened anaesthetized monkeys with 140-decibel noise and studied the neurological damage. Some wore monkey masks and sat in makeshift cages on the sidewalk.

My fumblings inflicted no cruelty; the mice were already dead. I had, I discovered, a personal problem with a mouse’s moment of death, but I didn’t think using mice for research was wrong. The protestors objected to specific acts of cruelty, but their protests overlaid an opposition to the exploitation of animals in general, mice and monkeys.

During the protests, one man climbed temporary scaffolding outside our lab windows, unfurled a banner, and stayed there for a week, four stories up. The police monitored him continuously from the office of my supervisor, away in Europe arranging our coming move. Juan said, “That guy wants to make a career as the Julia Butterfly Hill of animal research. He did the same kind of thing at Berkeley in the clock tower.” Animal activists will be even worse in Europe, the head of our department warned my supervisor after he returned.

After we had moved to Germany, no crowds of protestors appeared. Activists ignored the repeated loss of air conditioning in our laboratory’s new animal facility, even though some of our mice died from the heat. My co-workers became frustrated while breeding genetically engineered mice. The mothers would eat the newborn mice that were sickest, probably because they had weak bones from two mutated copies of the gene we were studying. I moved away from research on mice back to the realm of test tubes and cell culture dishes, but I remembered how apparently wasteful animal research can be.

A few years later, I was a summer intern at Science magazine in Cambridge, England. Shy and with few friends, I spent many evenings borrowing books from the Borders bookstore and lounging on the plump couches in its café. The staff broke my reverie one evening by gathering spare chairs for a book signing. The Cambridge Anti-Vivisection Society had invited a guest speaker.

Oh no, I thought. More examples like the deaf monkeys from UCSF. I picked up my backpack and prepared to leave the store. A man who had been reading across from me grumbled that he worked in a biochemistry lab and expressed disdain. I stuck around, curious to see what university scientists in the audience might ask the guest. And perhaps I was ready to hear what he had to say.

The speaker, a doctor from Alabama named Ray Greek, immediately dispelled my worries by saying: I’m not going to talk about cruelty or injustice. I’m going to talk about utility. Proponents of medical research using animals point to health benefits for humans. Let’s examine that claim.

Animals and humans are too different, he said. The results of animal experimentation send doctors trying new treatments in the clinic down blind alleys. He led the audience through a tour of different areas of medical research, arguing that using animals as models for humans was foolish.

I agreed with some of his points. The strengths of the most important enzymes that break down or modify drugs, a family of enzymes called cytochrome p450, vary greatly between humans. The Food and Drug Administration is encouraging research in pharmacogenomics: how our genetic differences determine how a drug will be processed in the body and how effective it will be. Given the wide variations between human individuals, it does seem misguided to depend on animal research to identify the winners from a large pool of possible drugs.

Mistakes with animal models have slowed down the development of drugs to combat hypertension and strokes. And cancer was another minefield, the doctor asserted. Researchers who study the steps required for cancer to develop have found that compared with mouse or monkey cells, human cells need a different set of genetic disruptions to unhinge their growth. He cited Richard Klausner, former head of the National Cancer Institute, who said: “The history of cancer research has been a history of curing cancer in the mouse… We have cured mice of cancer for decades — and it simply didn’t work in humans.”

When asked by a persistent heckler if he’d give drugs to a relative that hadn’t been screened for safety in animals, he presented long lists of clinically successful drugs that cause cancer or birth defects in mice. Epidemiology on people, cells on a plate, computer models, all would give better information. Half of the chemicals that cause cancer in rats don’t cause cancer in mice, he said; how can we make predictions about people? He reminded the audience more than once that the FDA requires animal tests only for the ability of chemicals to cause cancer or birth defects, not for efficacy. Animal testing is a way for drug companies to cover their rears and for researchers to keep their careers rolling, he said.

Many sympathetic analysts say that the drug industry now gets a poor return on its investment dollar. Greek described how executives at pharmaceutical companies had confided in him that they thought much of the money they spent on animal research was a waste.

I had more trouble with some of his other points. Dr. Greek said that research on AIDS and HIV using monkeys was senseless because they don’t get AIDS. I thought, “Oh yeah? Today I talked to a guy who just found a gene that if tweaked that could make a monkey vulnerable to AIDS. Won’t it be useful to know how that works?” More generally, I wondered, doesn’t it make sense to cross the models off the list that don’t have predictive power, but keep the ones that are useful?

I checked out from the library a book Dr. Greek and his wife, a veterinarian, had written. It attacks medical history with a more ideological bent than Greek seemed to take in his talk. It re-examines the stories of Pasteur, Koch, Fleming, and Sabin through the lens of animal rights. The idea that animals make a good model for humans misled them all, it claims.

It seemed like Dr. Greek was saying: in all these different areas, we know enough now to avoid using animals. I wasn’t sure how that left research into basic development, what my old laboratory was interested in. Scientists still don’t know what many genes do in an intact animal, and they don’t show up as natural mutations in humans. Is it cruel to create a mouse that suffocates after birth because it has weak bones or no ribs? Perhaps, but I know that scientists find those experimental examples invaluable in figuring out how bones grow or how the body pattern forms.

Although my research didn’t come to fruition, a colleague from the lab in San Francisco who stayed in California to work at Stanford found that the genes we worked on play a critical role in bone marrow stem cell renewal. Our favorite genes could make up part of a basic toolkit for replenishing and manipulating those elusive elixirs of the future.

My attitude towards animal research may in the future move from former President Clinton’s view of affirmative action: “Mend it, don’t end it,” to Justice O’Connor’s: (paraphrasing her 2003 opinion on University of Michigan Law School admissions) “It’s an evil that is temporarily necessary for the next couple of decades.” But I’m not there yet. Recent difficulties with experimental treatments like gene therapy — Jesse Gelsinger’s death in 1999 in Pennsylvania, for example — and fetal brain tissue transplantation show that in some areas researchers are still at the bottom of the learning curve. I believe patients considering complicated therapies like stem cells and gene therapy will want to know that researchers have exhaustively combed over the possible outcomes, using animals when no other model, at present, will do as well.