What the data actually shows
Attachment theory offers one of the clearest explanations. Building on John Bowlby's work and on Hazan and Shaver's research extending attachment to adult romantic relationships, this line of study describes broad styles — secure, anxious, avoidant — that form early and tend to carry into adulthood. These styles shape what feels normal in love, and they can draw people toward familiar dynamics: an anxiously attached person, for example, may repeatedly be drawn to more avoidant partners, recreating a familiar push-and-pull. The research describes tendencies, not destiny, and styles can shift over time and with different relationships.
Familiarity itself is part of the story. Decades of research on the mere-exposure effect find that people tend to prefer what they have encountered before, simply because it feels safe and easy to process. Applied to relationships, dynamics that resemble what you already know — even uncomfortable ones — can register as more comfortable than the unfamiliar, which nudges choices toward the familiar 'type.'
Then there is similarity. A large body of research on homophily finds people are generally drawn to partners who resemble them in values, background, education, and outlook — the opposite of the popular 'opposites attract' idea. And work by Park and MacDonald (2019) found that a person's ex-partners and current partner tend to resemble each other, suggesting a degree of consistency in the people we choose. Together these point to a 'type' that is partly about familiarity and partly about genuine, stable preference.
Why this feels different from how it actually is
It feels mysterious or fated because the pulls operate largely below conscious awareness. You do not experience 'I am seeking a familiar attachment dynamic'; you experience a spark, a sense of chemistry, an inexplicable draw. The underlying pattern is doing the steering while the felt experience is just attraction.
It can also feel like bad luck because the early phase of attraction is exactly when these pulls are strongest and least examined. What reads as instant connection is sometimes recognition — the comfortable familiarity of a dynamic you already know how to navigate, for better or worse.
And hindsight sharpens the pattern. Once a few relationships end in similar ways, it is natural to connect the dots into 'I always date the same type.' That story is often partly accurate, given the consistency the research finds — but it can also overstate things, since memory tends to emphasise the matches and quietly drop the relationships that did not fit the pattern.
A dynamic feeling 'right' early on is not strong evidence that it is healthy — only that it is recognisable.
What the research says to do about it
The most useful move the research supports is awareness itself. Because much of the pull is unconscious and familiarity-driven, naming your own patterns — the recurring dynamics, the traits you reliably gravitate toward, the early-relationship feelings you treat as 'chemistry' — makes them easier to question rather than simply act on. You cannot interrupt a pattern you cannot see.
Understanding your attachment tendencies can help. The attachment literature suggests that recognising an anxious or avoidant lean, and the kinds of partners it draws you toward, gives you a frame for noticing when familiarity is masquerading as compatibility. Attachment styles are not fixed; they can move toward security over time, often within relationships that consistently feel safe.
It also helps to separate familiar from good. Since the mere-exposure effect means familiarity reads as comfort, a dynamic feeling 'right' early on is not strong evidence that it is healthy — only that it is recognisable. Slowing down enough to evaluate a relationship on how it actually functions, rather than how immediately familiar it feels, is broadly consistent with what this research implies.
What the research says does not help
Treating your 'type' as fixed destiny does not help and is not what the evidence shows. These are patterns and tendencies — shaped by early experience and familiarity — not unchangeable wiring. Believing you are doomed to repeat them can become self-fulfilling.
Deliberately choosing the apparent 'opposite' of your usual partner, by itself, tends not to solve much. The research on homophily and on consistent partner traits suggests the underlying pulls are about familiarity and similarity, so a surface-level swap can leave the same dynamic intact — and the 'opposites attract' premise is itself largely a myth in the data.
Blaming yourself, or chalking it up to bad luck, both miss the mechanism. Repetition here is mostly an understandable product of attachment patterns, familiarity, and stable preferences — not a character defect and not random misfortune. Self-blame tends to obscure the pattern rather than reveal it.
These are patterns, not fate, and patterns tend to be the kind of thing awareness can change.
What this looks like in real life
When 'chemistry' is really recognition
The pulls operate below conscious awareness, so you don't experience 'I'm seeking a familiar attachment dynamic' — you experience a spark. An anxiously attached person, for instance, may feel an instant draw to a more avoidant partner, recreating a familiar push-and-pull. What reads as instant connection is sometimes just the comfort of a dynamic you already know how to navigate, for better or worse.
'The same person in a different body'
Someone looks back over a few relationships that ended in similar ways and concludes 'I always date the same type.' Given the consistency the research finds, that story is often partly accurate. But memory tends to emphasise the matches and quietly drop the relationships that didn't fit — so the pattern is usually real and recognisable, yet less absolute than it feels.
Real numbers in context
This is a domain of patterns and tendencies rather than crisp percentages, and it is worth holding the findings loosely. Attachment research (building on Bowlby; Hazan and Shaver) describes broad styles that form early and tend to persist, but with real individual variation and the capacity to change. The mere-exposure effect is one of the more robust findings in psychology, but its size in any single relationship choice is hard to pin down.
The most concrete relevant finding is Park and MacDonald's (2019) result that people's past and present partners tend to resemble one another on some characteristics — evidence that a consistent 'type' is real to a degree, not just a story we tell. Even there, the resemblance is partial rather than total, which fits the everyday experience of patterns that are recognisable but not absolute.