It really depends on what aspect of biological anthro you're dealign with.
Sociobiology tries to use genetics and biological needs to explain all human behavior. For example, females tend to marry males whoa re good providers because females are saddled with infant care and need males to provide food and protection. This is very theoretical but almost impossible to prove. So while it sounds scientific, it really isn't.
Some biological anthropologists are interested in evolutionary genetics, which helps explain why some human populations might be prone to certain diseases and disorders while others are not. For example, African-American populations descended from slaves who survived crossing the Atlantic. The ones who survived the crossing were the ones who needed less food and water. Slower metabolisms and higher salt retention rates in the body lent themselves to surviving food and water deprivations during the crossing. The genes for these characteristics were passed down to current African-Americans and this is why cardiovascular diseases/disorders and obesity occur so frequently in modern African-American populations than other ethnic groups. Evolutionary genetics is one of the most scientific applications of biological anthropology.
Forensic anthropology is also pretty scientific because it uses known patterns of bone damage and skeletal measurement data to identify how people were killed and help identify murder victims.
Paleo-anthropologists are the ones who study human evolution in general. They are fossil hunters and they look for archaeological evidence of early human and pre-human culture. This is the least scientific of the three examples I have given. Other than the dates these scientists get from scientific dating methods, the rest is guess work based on models derived from modern hunters and gatherer tribes or non-human animal species. These models really can't be projected back in time because modern people aren't Homo habilis and hyenas aren't Australopithecines, know what I mean?