When people think about relationships worth walking away from, the obvious candidates come to mind first, the clearly hostile, the openly unkind. Two studies measuring what actually happens in the body and in a room suggest the harder cases to spot, and possibly the more damaging ones, are the relationships that don’t look toxic at all on the surface.
The clearest evidence comes from a 2007 study by Julianne Holt-Lunstad, Bert Uchino, Timothy Smith, and Amber Hicks, published in Annals of Behavioral Medicine. The researchers brought 107 healthy adults into the lab, each with a same-sex friend, and randomly assigned each pair to be either a clearly supportive friendship or an “ambivalent” one, a friend rated as at least somewhat helpful and positive but also somewhat upsetting. When participants discussed a negative event with their friend, the ones with ambivalent friends showed the greatest spike in systolic blood pressure, higher than participants with consistently supportive friends. They also had higher resting heart rates and worse markers of physiological self-regulation around their ambivalent friends. Even sharing good news didn’t help, participants with ambivalent friends didn’t get the usual cardiovascular boost that sharing positive news typically produces. The mixed, unpredictable friend, not the openly negative one, was where the body showed the most strain.
A second study helps show that this kind of difficulty isn’t confined to close one-on-one friendships, it spreads through entire groups. In a 2002 study by Sigal Barsade, published in Administrative Science Quarterly, the researcher placed a trained confederate into 29 small groups, a total of 94 participants, instructing the confederate to display either a pleasant or unpleasant mood without anyone else in the group knowing an actor was present. Independent video coders and participants’ own self-reports both confirmed the confederate’s mood spread reliably through the rest of the group. Crucially, negative mood contagion was just as powerful as positive mood contagion, a single person’s bad mood infected a room every bit as efficiently as a good one did. The downstream effects were measurable too, groups exposed to the negative confederate showed less cooperation, more conflict, and worse perceived task performance.
Together, these two studies suggest that the people worth walking away from aren’t always identifiable by obvious hostility. Holt-Lunstad and colleagues’ research shows that a friend who is sometimes warm and sometimes upsetting can do more measurable physiological damage than a friend who is simply, consistently difficult, because the body never gets to relax into knowing what to expect. Barsade’s research shows that a single person’s negative affect, present in even one member of a small group, can quietly reshape how everyone else in that group feels, cooperates, and performs, without anyone consciously registering why. Between the two, the relationships most worth reconsidering may be the ones that feel confusing rather than the ones that feel obviously bad.
It’s worth being honest about what these two studies don’t establish. Holt-Lunstad and colleagues’ research measured cardiovascular reactivity during a single lab interaction, and while the pattern is a real and meaningful signal, it doesn’t capture what long-term exposure to an ambivalent relationship does to health over years. Barsade’s study involved short-term lab groups rather than long-standing real-world relationships, so while the contagion effect itself is well-documented, how quickly or strongly it applies to a lasting friendship or family relationship may look somewhat different.
Within those honest limits, the research offers a genuinely useful recalibration for anyone trying to figure out which relationships deserve real distance. It isn’t just the people who are clearly unkind. It’s often the ones who leave a person unsure what version of them will show up next, and the ones whose bad mood quietly reshapes a whole room without anyone noticing it happening.