At five days old the embryo is known as a blastocyst
Scientists have used gene analysis techniques to help identify the genetic profile of IVF embryos that result in a successful pregnancy.
The Australian researchers believe it could end the need to transfer more than one embryo into a woman's womb to maximise the chances of success.
If only viable embryos were selected, there would be no chance of potentially risky multiple pregnancies.
The study appears in the journal Human Reproduction.
Following IVF treatment embryos are implanted in the women's womb when they reach the blastocyst stage, at about five days.
However, it is difficult to spot which embryos are healthy and which are likely to fail.
Clinic staff tend to make an assessment based on examination of blastocyst shape under the microscope.
Consequently many clinics transfer more than one embryo to maximise the chance of success. In the UK clinics are supposed only to transfer a maximum of two, but in some cases they opt for three.
This raises the odds of a multiple pregnancy, with raised risk to both the mother and her babies.
However, when multiple embryos are transferred, it has been impossible to work out which were the ones that developed into a successful pregnancy - making it difficult to develop criteria for identifying viable blastocysts.
The latest work by Monash University set out to solve that problem, and to identify not only which blastocysts were developing successfully following transfer, but also to identify the genetic blueprint which gave them that capacity.
In theory, this would allow IVF clinics to screen out those embryos which do not have the right genes before implantation.
The researchers recruited 48 women undergoing IVF. When the women's embryos reached the blastocyst stage, the researchers removed between eight and 20 cells from a cell layer known as the trophectoderm.
These cells were then analysed using sophisticated genetic amplification techniques.
All the women in the study then at had at least one of their blastocysts transferred to their womb.
When the babies were born, blood from the umbilical cord or swabs of cheek cells were taken and stored.
The researchers used DNA fingerprinting on these samples to match them with the DNA obtained from the blastocyst cells, enabling them to identify which embryos had successfully developed to full term.
Then they used genetic techniques to find out which genes were expressed in the viable blastocysts.
The work is still continuing, but they have already identified groups of key genes, and hope to refine the process further.
Researcher Dr Gayle Jones said: "The ability to select the single most viable embryo from within a cohort available for transfer will revolutionise the practice of IVF, not only improving pregnancy rates but eliminating multiple pregnancies and the attendant complications."
Dr Simon Fishel, director of the CARE Fertility Centre in Nottingham, said the technique showed promise.
He said it would be even more useful if a single cell (blastomere) sample could be taken from the embryo when it was about the eight cell stage.
This would allow doctors two days to get back the results of the genetic analysis, and to implant the embryo into the womb while still fresh, without the need to freeze it, which can cause damage.
Dr Fishel's team has developed a method to assess the chromosomal make up of embryos. He said combining the two techniques could be a powerful way to ensure the selection of viable embryos.
He said: "The golden goal for all practitioners is one embryo, one baby."
Dr Allan Pacey, senior lecturer at the University of Sheffield and honorary secretary of the British Fertility Society, said: "As we move increasingly toward elective single embryo transfer for as many patients as possible, it will become increasingly important to be able to select which embryos are the most likely to lead to pregnancies."